diff --git a/README.md b/README.md
index 97c75fd..a7cfd79 100644
--- a/README.md
+++ b/README.md
@@ -20,7 +20,7 @@
 <a href="https://giphy.com/gifs/r3GaZz7H1H6zpuIvPI/fullscreen"><img src="https://media.giphy.com/media/13oe6zo6b2B7CdsOac/giphy.gif"></a>    <a href="https://giphy.com/gifs/FjzrGJxsOzZAXaW7Vi/fullscreen"><img src="https://media.giphy.com/media/4JLERHxOEgH0tt5DZO/giphy.gif"></a>
 </p>
 
-This branch contains the pytorch implementation of <a href="https://arxiv.org/abs/2204.12484">ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation</a>. It obtains 81.1 AP on MS COCO Keypoint test-dev set.
+This branch contains the pytorch implementation of <a href="https://arxiv.org/abs/2204.12484">ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation</a> and <a href="https://arxiv.org/abs/2212.04246">ViTPose+: Vision Transformer Foundation Model for Generic Body Pose Estimation</a>. It obtains 81.1 AP on MS COCO Keypoint test-dev set.
 
 <img src="figures/Throughput.png" class="left" width='80%'>
 
@@ -28,6 +28,10 @@ This branch contains the pytorch implementation of <a href="https://arxiv.org/ab
 
 - Integrated into [Huggingface Spaces 🤗](https://huggingface.co/spaces) using [Gradio](https://github.com/gradio-app/gradio). Try out the Web Demo for video: [![Hugging Face Spaces](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Spaces-blue)](https://huggingface.co/spaces/hysts/ViTPose_video) and images [![Hugging Face Spaces](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Spaces-blue)](https://huggingface.co/spaces/Gradio-Blocks/ViTPose)
 
+## Pre-trained model
+
+The pre-trained models can be found in the corresponding entry. The small size MAE pre-trained model can be found in [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccZeiFjh4DJ7gjYyg?e=iTMdMq).
+
 ## Results from this repo on MS COCO val set (single-task training)
 
 Using detection results from a detector that obtains 56 mAP on person. The configs here are for both training and test.
@@ -36,6 +40,7 @@ Using detection results from a detector that obtains 56 mAP on person. The confi
 
 | Model | Pretrain | Resolution | AP | AR | config | log | weight |
 | :----: | :----: | :----: | :----: | :----: | :----: | :----: | :----: |
+| ViTPose-S | MAE | 256x192 | 73.8 | 79.2 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_small_coco_256x192.py) | [log](https://1drv.ms/u/s!AimBgYV7JjTlgcchdNXBAh7ClS14pA?e=dKXmJ6) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccifT1XlGRatxg3vw?e=9wz7BY) |
 | ViTPose-B | MAE | 256x192 | 75.8 | 81.1 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_base_coco_256x192.py) | [log](logs/vitpose-b.log.json) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgSMjp1_NrV3VRSmK?e=Q1uZKs) |
 | ViTPose-L | MAE | 256x192 | 78.3 | 83.5 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_large_coco_256x192.py) | [log](logs/vitpose-l.log.json) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgSd9k_kuktPtiP4F?e=K7DGYT) |
 | ViTPose-H | MAE | 256x192 | 79.1 | 84.1 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_huge_coco_256x192.py) | [log](logs/vitpose-h.log.json) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgShLMI-kkmvNfF_h?e=dEhGHe) |
@@ -44,6 +49,7 @@ Using detection results from a detector that obtains 56 mAP on person. The confi
 
 | Model | Pretrain | Resolution | AP | AR | config | log | weight |
 | :----: | :----: | :----: | :----: | :----: | :----: | :----: | :----: |
+| ViTPose-S | MAE | 256x192 | 73.5 | 78.9 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_small_simple_coco_256x192.py) | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccfkqELJqE67kpRtw?e=InSjJP) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccgb_50jIgiYkHvdw?e=D7RbH2) |
 | ViTPose-B | MAE | 256x192 | 75.5 | 80.9 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_base_simple_coco_256x192.py) | [log](logs/vitpose-b-simple.log.json) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgSRPKrD5PmDRiv0R?e=jifvOe) |
 | ViTPose-L | MAE | 256x192 | 78.2 | 83.4 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_large_simple_coco_256x192.py) | [log](logs/vitpose-l-simple.log.json) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgSVS6DP2LmKwZ3sm?e=MmCvDT) |
 | ViTPose-H | MAE | 256x192 | 78.9 | 84.0 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_huge_simple_coco_256x192.py) | [log](logs/vitpose-h-simple.log.json) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgSbHyN2mjh2n2LyG?e=y0FgMK) |
@@ -52,72 +58,129 @@ Using detection results from a detector that obtains 56 mAP on person. The confi
 
 **Note** \* There may exist duplicate images in the crowdpose training set and the validation images in other datasets, as discussed in [issue #24](https://github.com/ViTAE-Transformer/ViTPose/issues/24). Please be careful when using these models for evaluation. We provide the results without the crowpose dataset for reference.
 
-### Results on MS COCO val set
+### Human datasets (MS COCO, AIC, MPII, CrowdPose)
+> Results on MS COCO val set
 
 Using detection results from a detector that obtains 56 mAP on person. Note the configs here are only for evaluation.
 
 | Model | Dataset | Resolution | AP | AR | config | weight |
 | :----: | :----: | :----: | :----: | :----: | :----: | :----: |
-| ViTPose-B | COCO+AIC+MPII | 256x192 | 77.1 | 82.2 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_base_coco_256x192.py)  | Coming Soon |
-| ViTPose-L | COCO+AIC+MPII | 256x192 | 78.7 | 83.8 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_large_coco_256x192.py) | Coming Soon |
-| ViTPose-H | COCO+AIC+MPII | 256x192 | 79.5 | 84.5 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_huge_coco_256x192.py) | Coming Soon |
+| ViTPose-B | COCO+AIC+MPII | 256x192 | 77.1 | 82.2 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_base_coco_256x192.py)  | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgcccwaTZ8xCFFM3Sjg?e=chmiK5) |
+| ViTPose-L | COCO+AIC+MPII | 256x192 | 78.7 | 83.8 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_large_coco_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccdOLQqSo6E87GfMw?e=TEurgW) |
+| ViTPose-H | COCO+AIC+MPII | 256x192 | 79.5 | 84.5 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_huge_coco_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccmHofkmfJDQDukVw?e=gRK224) |
+| ViTPose-G | COCO+AIC+MPII | 576x432 | 81.0 | 85.6 | | |
+| **ViTPose+-S** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 75.8 | 82.6 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_small_coco_256x192.py)  | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccqO1JBHtBjNaeCbQ?e=ZN5NSz) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccrwORr61gT9E4n8g?e=kz9sz5) |
+| **ViTPose+-B** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 77.0 | 82.6 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_base_coco_256x192.py)  | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccjj9lgPTlkGT1HTw?e=OlS5zv) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgcckRZk1bIAuRa_E1w?e=ylDB2G) |
+| **ViTPose+-L** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 78.6 | 84.1 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_large_coco_256x192.py) | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccp7HJf4QMeQQpeyA?e=JagPNt) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccs1SNFUGSTsmRJ8w?e=a9zKwZ) |
+| **ViTPose+-H** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 79.4 | 84.8 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_huge_coco_256x192.py) | [log](https://1drv.ms/u/s!AimBgYV7JjTlgcclxZOlwRJdqpIIjA?e=nFQgVC) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccoXv8rCUgVe7oD9Q?e=ZBw6gR) |
 | ViTPose-B* | COCO+AIC+MPII+CrowdPose | 256x192 | 77.5 | 82.6 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_base_coco_256x192.py)  |[Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgSrlMB093JzJtqq-?e=Jr5S3R) |
 | ViTPose-L* | COCO+AIC+MPII+CrowdPose | 256x192 | 79.1 | 84.1 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_large_coco_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgTBm3dCVmBUbHYT6?e=fHUrTq) |
 | ViTPose-H* | COCO+AIC+MPII+CrowdPose | 256x192 | 79.8 | 84.8 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_huge_coco_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgS5rLeRAJiWobCdh?e=41GsDd) |
-| ViTPose-G* | COCO+AIC+MPII+CrowdPose | 576x432 | 81.0 | 85.6 | | |
 
-### Results on OCHuman test set
+
+> Results on OCHuman test set
 
 Using groundtruth bounding boxes. Note the configs here are only for evaluation.
 
 | Model | Dataset | Resolution | AP | AR | config | weight |
 | :----: | :----: | :----: | :----: | :----: | :----: | :----: |
-| ViTPose-B | COCO+AIC+MPII | 256x192 | 88.0 | 89.6 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_base_ochuman_256x192.py)  | Coming Soon |
-| ViTPose-L | COCO+AIC+MPII | 256x192 | 90.9 | 92.2 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_large_ochuman_256x192.py) | Coming Soon |
-| ViTPose-H | COCO+AIC+MPII | 256x192 | 90.9 | 92.3 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_huge_ochuman_256x192.py) | Coming Soon |
+| ViTPose-B | COCO+AIC+MPII | 256x192 | 88.0 | 89.6 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_base_ochuman_256x192.py)  | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgcccwaTZ8xCFFM3Sjg?e=chmiK5) |
+| ViTPose-L | COCO+AIC+MPII | 256x192 | 90.9 | 92.2 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_large_ochuman_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccdOLQqSo6E87GfMw?e=TEurgW) |
+| ViTPose-H | COCO+AIC+MPII | 256x192 | 90.9 | 92.3 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_huge_ochuman_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccmHofkmfJDQDukVw?e=gRK224) |
+| ViTPose-G | COCO+AIC+MPII | 576x432 | 93.3 | 94.3 | | |
+| **ViTPose+-S** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 78.4 | 80.6 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_small_ochuman_256x192.py)  | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccqO1JBHtBjNaeCbQ?e=ZN5NSz) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccrwORr61gT9E4n8g?e=kz9sz5) |
+| **ViTPose+-B** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 82.6 | 84.8 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_base_ochuman_256x192.py)  | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccjj9lgPTlkGT1HTw?e=OlS5zv) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgcckRZk1bIAuRa_E1w?e=ylDB2G) |
+| **ViTPose+-L** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 85.7 | 87.5 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_large_ochuman_256x192.py) | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccp7HJf4QMeQQpeyA?e=JagPNt) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccs1SNFUGSTsmRJ8w?e=a9zKwZ) |
+| **ViTPose+-H** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 85.7 | 87.4 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_huge_ochuman_256x192.py) | [log](https://1drv.ms/u/s!AimBgYV7JjTlgcclxZOlwRJdqpIIjA?e=nFQgVC) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccoXv8rCUgVe7oD9Q?e=ZBw6gR) |
 | ViTPose-B* | COCO+AIC+MPII+CrowdPose | 256x192 | 88.2 | 90.0 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_base_ochuman_256x192.py)  |[Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgSrlMB093JzJtqq-?e=Jr5S3R) |
 | ViTPose-L* | COCO+AIC+MPII+CrowdPose | 256x192 | 91.5 | 92.8 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_large_ochuman_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgTBm3dCVmBUbHYT6?e=fHUrTq) |
 | ViTPose-H* | COCO+AIC+MPII+CrowdPose | 256x192 | 91.6 | 92.8 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_huge_ochuman_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgS5rLeRAJiWobCdh?e=41GsDd) |
-| ViTPose-G* | COCO+AIC+MPII+CrowdPose | 576x432 | 93.3 | 94.3 | | |
 
-### Results on MPII val set
+
+> Results on MPII val set
 
 Using groundtruth bounding boxes. Note the configs here are only for evaluation. The metric is PCKh.
 
 | Model | Dataset | Resolution | Mean | config | weight |
 | :----: | :----: | :----: | :----: | :----: | :----: |
-| ViTPose-B | COCO+AIC+MPII | 256x192 | 93.3 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/mpii/ViTPose_base_mpii_256x192.py)  |Coming Soon |
-| ViTPose-L | COCO+AIC+MPII | 256x192 | 94.0 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/mpii/ViTPose_large_mpii_256x192.py) | Coming Soon |
-| ViTPose-H | COCO+AIC+MPII | 256x192 | 94.1 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/mpii/ViTPose_huge_mpii_256x192.py) | Coming Soon |
+| ViTPose-B | COCO+AIC+MPII | 256x192 | 93.3 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/mpii/ViTPose_base_mpii_256x192.py)  | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgcccwaTZ8xCFFM3Sjg?e=chmiK5) |
+| ViTPose-L | COCO+AIC+MPII | 256x192 | 94.0 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/mpii/ViTPose_large_mpii_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccdOLQqSo6E87GfMw?e=TEurgW) |
+| ViTPose-H | COCO+AIC+MPII | 256x192 | 94.1 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/mpii/ViTPose_huge_mpii_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccmHofkmfJDQDukVw?e=gRK224) |
+| ViTPose-G | COCO+AIC+MPII | 576x432 | 94.3 | | |
+| **ViTPose+-S** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 92.7 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/mpii/ViTPose_small_mpii_256x192.py)  | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccqO1JBHtBjNaeCbQ?e=ZN5NSz) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccrwORr61gT9E4n8g?e=kz9sz5) |
+| **ViTPose+-B** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 92.8 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/mpii/ViTPose_base_mpii_256x192.py)  | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccjj9lgPTlkGT1HTw?e=OlS5zv) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgcckRZk1bIAuRa_E1w?e=ylDB2G) |
+| **ViTPose+-L** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 94.0 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/mpii/ViTPose_large_mpii_256x192.py) | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccp7HJf4QMeQQpeyA?e=JagPNt) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccs1SNFUGSTsmRJ8w?e=a9zKwZ) |
+| **ViTPose+-H** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 94.2 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/mpii/ViTPose_huge_mpii_256x192.py) | [log](https://1drv.ms/u/s!AimBgYV7JjTlgcclxZOlwRJdqpIIjA?e=nFQgVC) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccoXv8rCUgVe7oD9Q?e=ZBw6gR) |
 | ViTPose-B* | COCO+AIC+MPII+CrowdPose | 256x192 | 93.4 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/mpii/ViTPose_base_mpii_256x192.py)  |[Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgSy_OSEm906wd2LB?e=GOSg14) |
 | ViTPose-L* | COCO+AIC+MPII+CrowdPose | 256x192 | 93.9 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/mpii/ViTPose_large_mpii_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgTM32I6Kpjr-esl6?e=qvh0Yl) |
 | ViTPose-H* | COCO+AIC+MPII+CrowdPose | 256x192 | 94.1 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/mpii/ViTPose_huge_mpii_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgTT90XEQBKy-scIH?e=D2WhTS) |
-| ViTPose-G* | COCO+AIC+MPII+CrowdPose | 576x432 | 94.3 | | |
 
-### Results on AI Challenger test set
+
+> Results on AI Challenger test set
 
 Using groundtruth bounding boxes. Note the configs here are only for evaluation.
 
 | Model | Dataset | Resolution | AP | AR | config | weight |
 | :----: | :----: | :----: | :----: | :----: | :----: | :----: |
-| ViTPose-B | COCO+AIC+MPII | 256x192 | 32.0 | 36.3 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/aic/ViTPose_base_aic_256x192.py)  | Coming Soon |
-| ViTPose-L | COCO+AIC+MPII | 256x192 | 34.5 | 39.0 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/aic/ViTPose_large_aic_256x192.py) | Coming Soon |
-| ViTPose-H | COCO+AIC+MPII | 256x192 | 35.4 | 39.9 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/aic/ViTPose_huge_aic_256x192.py) | Coming Soon |
+| ViTPose-B | COCO+AIC+MPII | 256x192 | 32.0 | 36.3 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/aic/ViTPose_base_aic_256x192.py)  | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgcccwaTZ8xCFFM3Sjg?e=chmiK5) |
+| ViTPose-L | COCO+AIC+MPII | 256x192 | 34.5 | 39.0 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/aic/ViTPose_large_aic_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccdOLQqSo6E87GfMw?e=TEurgW) |
+| ViTPose-H | COCO+AIC+MPII | 256x192 | 35.4 | 39.9 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/aic/ViTPose_huge_aic_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccmHofkmfJDQDukVw?e=gRK224) |
+| ViTPose-G | COCO+AIC+MPII | 576x432 | 43.2 | 47.1 | | |
+| **ViTPose+-S** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 29.7 | 34.3 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_small_ochuman_256x192.py)  | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccqO1JBHtBjNaeCbQ?e=ZN5NSz) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccrwORr61gT9E4n8g?e=kz9sz5) |
+| **ViTPose+-B** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 31.8 | 36.3 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_base_ochuman_256x192.py)  | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccjj9lgPTlkGT1HTw?e=OlS5zv) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgcckRZk1bIAuRa_E1w?e=ylDB2G) |
+| **ViTPose+-L** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 34.3 | 38.9 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_large_ochuman_256x192.py) | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccp7HJf4QMeQQpeyA?e=JagPNt) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccs1SNFUGSTsmRJ8w?e=a9zKwZ) |
+| **ViTPose+-H** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 34.8 | 39.1 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_huge_ochuman_256x192.py) | [log](https://1drv.ms/u/s!AimBgYV7JjTlgcclxZOlwRJdqpIIjA?e=nFQgVC) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccoXv8rCUgVe7oD9Q?e=ZBw6gR) |70707
 | ViTPose-B* | COCO+AIC+MPII+CrowdPose | 256x192 | 31.9 | 36.3 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/aic/ViTPose_base_aic_256x192.py)  |[Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgSlvdVaXTC92SHYH?e=j7iqcp) |
 | ViTPose-L* | COCO+AIC+MPII+CrowdPose | 256x192 | 34.6 | 39.0 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/aic/ViTPose_large_aic_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgTF06FX3FSAm0MOH?e=rYts9F) |
 | ViTPose-H* | COCO+AIC+MPII+CrowdPose | 256x192 | 35.3 | 39.8 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/aic/ViTPose_huge_aic_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgS1MRmb2mcow_K04?e=q9jPab) |
-| ViTPose-G* | COCO+AIC+MPII+CrowdPose | 576x432 | 43.2 | 47.1 | | |
 
-### Results on CrowdPose test set
+> Results on CrowdPose test set
 
 Using YOLOv3 human detector. Note the configs here are only for evaluation.
 
 | Model | Dataset | Resolution | AP | AP(H) | config | weight |
 | :----: | :----: | :----: | :----: | :----: | :----: | :----: |
-| ViTPose-B | COCO+AIC+MPII+CrowdPose | 256x192 | 74.7 | 63.3 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/crowdpose/ViTPose_base_crowdpose_256x192.py)  |[Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgStrrCb91cPlaxJx?e=6Xobo6) |
-| ViTPose-L | COCO+AIC+MPII+CrowdPose | 256x192 | 76.6 | 65.9 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/crowdpose/ViTPose_large_crowdpose_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgTK3dug-r7c6GFyu?e=1ZBpEG) |
-| ViTPose-H | COCO+AIC+MPII+CrowdPose | 256x192 | 76.3 | 65.6 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/crowdpose/ViTPose_huge_crowdpose_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgS-oAvEV4MTD--Xr?e=EeW2Fu) |
-| ViTPose-G | COCO+AIC+MPII+CrowdPose | 576x432 | 78.3 | 67.9 | | |
+| ViTPose-B* | COCO+AIC+MPII+CrowdPose | 256x192 | 74.7 | 63.3 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/crowdpose/ViTPose_base_crowdpose_256x192.py)  |[Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgStrrCb91cPlaxJx?e=6Xobo6) |
+| ViTPose-L* | COCO+AIC+MPII+CrowdPose | 256x192 | 76.6 | 65.9 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/crowdpose/ViTPose_large_crowdpose_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgTK3dug-r7c6GFyu?e=1ZBpEG) |
+| ViTPose-H* | COCO+AIC+MPII+CrowdPose | 256x192 | 76.3 | 65.6 | [config](configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/crowdpose/ViTPose_huge_crowdpose_256x192.py) | [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgS-oAvEV4MTD--Xr?e=EeW2Fu) |
+
+### Animal datasets (AP10K, APT36K)
+
+> Results on AP-10K test set
+
+| Model | Dataset | Resolution | AP | config | weight |
+| :----: | :----: | :----: | :----: | :----: | :----: |
+| **ViTPose+-S** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 71.4 | [config](configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/ap10k/ViTPose_small_ap10k_256x192.py)  | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccqO1JBHtBjNaeCbQ?e=ZN5NSz) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccrwORr61gT9E4n8g?e=kz9sz5) |
+| **ViTPose+-B** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 74.5 | [config](configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/ap10k/ViTPose_base_ap10k_256x192.py)  | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccjj9lgPTlkGT1HTw?e=OlS5zv) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgcckRZk1bIAuRa_E1w?e=ylDB2G) |
+| **ViTPose+-L** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 80.4 | [config](configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/ap10k/ViTPose_large_ap10k_256x192.py) | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccp7HJf4QMeQQpeyA?e=JagPNt) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccs1SNFUGSTsmRJ8w?e=a9zKwZ) |
+| **ViTPose+-H** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 82.4 | [config](configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/ap10k/ViTPose_huge_ap10k_256x192.py) | [log](https://1drv.ms/u/s!AimBgYV7JjTlgcclxZOlwRJdqpIIjA?e=nFQgVC) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccoXv8rCUgVe7oD9Q?e=ZBw6gR) |
+
+> Results on APT-36K val set
+
+| Model | Dataset | Resolution | AP | config | weight |
+| :----: | :----: | :----: | :----: | :----: | :----: |
+| **ViTPose+-S** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 74.2 | [config](configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/apt36k/ViTPose_small_apt36k_256x192.py)  | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccqO1JBHtBjNaeCbQ?e=ZN5NSz) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccrwORr61gT9E4n8g?e=kz9sz5) |
+| **ViTPose+-B** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 75.9 | [config](configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/apt36k/ViTPose_base_apt36k_256x192.py)  | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccjj9lgPTlkGT1HTw?e=OlS5zv) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgcckRZk1bIAuRa_E1w?e=ylDB2G) |
+| **ViTPose+-L** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 80.8 | [config](configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/apt36k/ViTPose_large_apt36k_256x192.py) | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccp7HJf4QMeQQpeyA?e=JagPNt) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccs1SNFUGSTsmRJ8w?e=a9zKwZ) |
+| **ViTPose+-H** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 82.3 | [config](configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/apt36k/ViTPose_huge_apt36k_256x192.py) | [log](https://1drv.ms/u/s!AimBgYV7JjTlgcclxZOlwRJdqpIIjA?e=nFQgVC) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccoXv8rCUgVe7oD9Q?e=ZBw6gR) |
+
+### WholeBody dataset
+
+| Model | Dataset | Resolution | AP | config | weight |
+| :----: | :----: | :----: | :----: | :----: | :----: |
+| **ViTPose+-S** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 54.4 | [config](configs/wholebody/2d_kpt_sview_rgb_img/topdown_heatmap/coco-wholebody/ViTPose_small_wholebody_256x192.py)  | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccqO1JBHtBjNaeCbQ?e=ZN5NSz) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccrwORr61gT9E4n8g?e=kz9sz5) |
+| **ViTPose+-B** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 57.4 | [config](cconfigs/wholebody/2d_kpt_sview_rgb_img/topdown_heatmap/coco-wholebody/ViTPose_base_wholebody_256x192.py)  | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccjj9lgPTlkGT1HTw?e=OlS5zv) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgcckRZk1bIAuRa_E1w?e=ylDB2G) |
+| **ViTPose+-L** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 60.6 | [config](configs/wholebody/2d_kpt_sview_rgb_img/topdown_heatmap/coco-wholebody/ViTPose_large_wholebody_256x192.py) | [log](https://1drv.ms/u/s!AimBgYV7JjTlgccp7HJf4QMeQQpeyA?e=JagPNt) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccs1SNFUGSTsmRJ8w?e=a9zKwZ) |
+| **ViTPose+-H** | COCO+AIC+MPII+AP10K+APT36K+WholeBody | 256x192 | 61.2 | [config](configs/wholebody/2d_kpt_sview_rgb_img/topdown_heatmap/coco-wholebody/ViTPose_huge_wholebody_256x192.py) | [log](https://1drv.ms/u/s!AimBgYV7JjTlgcclxZOlwRJdqpIIjA?e=nFQgVC) \| [Onedrive](https://1drv.ms/u/s!AimBgYV7JjTlgccoXv8rCUgVe7oD9Q?e=ZBw6gR) |
+
+### Transfer results on the hand dataset (InterHand2.6M)
+
+| Model | Dataset | Resolution | AUC | config | weight |
+| :----: | :----: | :----: | :----: | :----: | :----: |
+| **ViTPose+-S** | COCO+AIC+MPII+WholeBody | 256x192 | 86.5 | [config](configs/hand/2d_kpt_sview_rgb_img/topdown_heatmap/interhand2d/ViTPose_small_interhand2d_all_256x192.py)  | Coming Soon |
+| **ViTPose+-B** | COCO+AIC+MPII+WholeBody | 256x192 | 87.0 | [config](configs/hand/2d_kpt_sview_rgb_img/topdown_heatmap/interhand2d/ViTPose_base_interhand2d_all_256x192.py)  | Coming Soon |
+| **ViTPose+-L** | COCO+AIC+MPII+WholeBody | 256x192 | 87.5 | [config](configs/hand/2d_kpt_sview_rgb_img/topdown_heatmap/interhand2d/ViTPose_large_interhand2d_all_256x192.py) | Coming Soon |
+| **ViTPose+-H** | COCO+AIC+MPII+WholeBody | 256x192 | 87.6 | [config](configs/hand/2d_kpt_sview_rgb_img/topdown_heatmap/interhand2d/ViTPose_huge_interhand2d_all_256x192.py) | Coming Soon |
 
 ## Updates
 
@@ -164,6 +227,12 @@ To test the pretrained models performance, please run
 bash tools/dist_test.sh <Config PATH> <Checkpoint PATH> <NUM GPUs>
 ```
 
+For ViTPose+ pre-trained models, please first re-organize the pre-trained weights using
+
+```bash
+python tools/model_split.py --source <Pretrained PATH>
+```
+
 ## Todo
 
 This repo current contains modifications including:
@@ -172,20 +241,33 @@ This repo current contains modifications including:
 
 - [x] More models with SOTA results
 
-- [ ] Upload multi-task training config
+- [x] Upload multi-task training config
 
 ## Acknowledge
 We acknowledge the excellent implementation from [mmpose](https://github.com/open-mmlab/mmdetection) and [MAE](https://github.com/facebookresearch/mae).
 
 ## Citing ViTPose
+
+For ViTPose
+
+```
+@inproceedings{
+  xu2022vitpose,
+  title={Vi{TP}ose: Simple Vision Transformer Baselines for Human Pose Estimation},
+  author={Yufei Xu and Jing Zhang and Qiming Zhang and Dacheng Tao},
+  booktitle={Advances in Neural Information Processing Systems},
+  year={2022},
+}
 ```
-@misc{xu2022vitpose,
-      title={ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation}, 
-      author={Yufei Xu and Jing Zhang and Qiming Zhang and Dacheng Tao},
-      year={2022},
-      eprint={2204.12484},
-      archivePrefix={arXiv},
-      primaryClass={cs.CV}
+
+For ViTPose+
+
+```
+@article{xu2022vitpose+,
+  title={ViTPose+: Vision Transformer Foundation Model for Generic Body Pose Estimation},
+  author={Xu, Yufei and Zhang, Jing and Zhang, Qiming and Tao, Dacheng},
+  journal={arXiv preprint arXiv:2212.04246},
+  year={2022}
 }
 ```
 
diff --git a/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/ap10k/ViTPose_base_ap10k_256x192.py b/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/ap10k/ViTPose_base_ap10k_256x192.py
new file mode 100644
index 0000000..bd5daf5
--- /dev/null
+++ b/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/ap10k/ViTPose_base_ap10k_256x192.py
@@ -0,0 +1,157 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/ap10k.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+log_config = dict(
+    interval=1,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=17,
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=768,
+        depth=12,
+        num_heads=12,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=768,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=True,
+    det_bbox_thr=0.0,
+    bbox_file='',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/apt36k'
+data = dict(
+    samples_per_gpu=32,
+    workers_per_gpu=4,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/train_annotations_1.json',
+        img_prefix=f'{data_root}/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/val_annotations_1.json',
+        img_prefix=f'{data_root}/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/val_annotations_1.json',
+        img_prefix=f'{data_root}/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/ap10k/ViTPose_huge_ap10k_256x192.py b/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/ap10k/ViTPose_huge_ap10k_256x192.py
new file mode 100644
index 0000000..1d2f8ab
--- /dev/null
+++ b/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/ap10k/ViTPose_huge_ap10k_256x192.py
@@ -0,0 +1,157 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/ap10k.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+log_config = dict(
+    interval=1,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=17,
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=1280,
+        depth=32,
+        num_heads=16,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=1280,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=True,
+    det_bbox_thr=0.0,
+    bbox_file='',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/ap10k'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=4,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/ap10k-train-split1.json',
+        img_prefix=f'{data_root}/data/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/ap10k-val-split1.json',
+        img_prefix=f'{data_root}/data/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/ap10k-test-split1.json',
+        img_prefix=f'{data_root}/data/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/ap10k/ViTPose_large_ap10k_256x192.py b/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/ap10k/ViTPose_large_ap10k_256x192.py
new file mode 100644
index 0000000..6e44c27
--- /dev/null
+++ b/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/ap10k/ViTPose_large_ap10k_256x192.py
@@ -0,0 +1,157 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/ap10k.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+log_config = dict(
+    interval=1,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=17,
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=1024,
+        depth=24,
+        num_heads=16,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=1024,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=True,
+    det_bbox_thr=0.0,
+    bbox_file='',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/ap10k'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=4,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/ap10k-train-split1.json',
+        img_prefix=f'{data_root}/data/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/ap10k-val-split1.json',
+        img_prefix=f'{data_root}/data/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/ap10k-test-split1.json',
+        img_prefix=f'{data_root}/data/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/ap10k/ViTPose_small_ap10k_256x192.py b/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/ap10k/ViTPose_small_ap10k_256x192.py
new file mode 100644
index 0000000..3c3f2b9
--- /dev/null
+++ b/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/ap10k/ViTPose_small_ap10k_256x192.py
@@ -0,0 +1,157 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/ap10k.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+log_config = dict(
+    interval=1,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=17,
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=384,
+        depth=12,
+        num_heads=12,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=384,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=True,
+    det_bbox_thr=0.0,
+    bbox_file='',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/ap10k'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=4,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/ap10k-train-split1.json',
+        img_prefix=f'{data_root}/data/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/ap10k-val-split1.json',
+        img_prefix=f'{data_root}/data/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/ap10k-test-split1.json',
+        img_prefix=f'{data_root}/data/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/apt36k/ViTPose_base_apt36k_256x192.py b/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/apt36k/ViTPose_base_apt36k_256x192.py
new file mode 100644
index 0000000..e3aa5d4
--- /dev/null
+++ b/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/apt36k/ViTPose_base_apt36k_256x192.py
@@ -0,0 +1,157 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/ap10k.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+log_config = dict(
+    interval=1,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=17,
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=768,
+        depth=12,
+        num_heads=12,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=768,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=True,
+    det_bbox_thr=0.0,
+    bbox_file='',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/ap10k'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=4,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/ap10k-train-split1.json',
+        img_prefix=f'{data_root}/data/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/ap10k-val-split1.json',
+        img_prefix=f'{data_root}/data/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/ap10k-test-split1.json',
+        img_prefix=f'{data_root}/data/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/apt36k/ViTPose_huge_ap36k_256x192.py b/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/apt36k/ViTPose_huge_ap36k_256x192.py
new file mode 100644
index 0000000..0562e79
--- /dev/null
+++ b/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/apt36k/ViTPose_huge_ap36k_256x192.py
@@ -0,0 +1,157 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/ap10k.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+log_config = dict(
+    interval=1,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=17,
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=1280,
+        depth=32,
+        num_heads=16,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=1280,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=True,
+    det_bbox_thr=0.0,
+    bbox_file='',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/apt36k'
+data = dict(
+    samples_per_gpu=32,
+    workers_per_gpu=4,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/train_annotations_1.json',
+        img_prefix=f'{data_root}/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/val_annotations_1.json',
+        img_prefix=f'{data_root}/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/val_annotations_1.json',
+        img_prefix=f'{data_root}/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/apt36k/ViTPose_large_apt36k_256x192.py b/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/apt36k/ViTPose_large_apt36k_256x192.py
new file mode 100644
index 0000000..d4ae268
--- /dev/null
+++ b/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/apt36k/ViTPose_large_apt36k_256x192.py
@@ -0,0 +1,157 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/ap10k.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+log_config = dict(
+    interval=1,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=17,
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=1024,
+        depth=24,
+        num_heads=16,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=1024,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=True,
+    det_bbox_thr=0.0,
+    bbox_file='',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/apt36k'
+data = dict(
+    samples_per_gpu=32,
+    workers_per_gpu=4,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/train_annotations_1.json',
+        img_prefix=f'{data_root}/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/val_annotations_1.json',
+        img_prefix=f'{data_root}/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/val_annotations_1.json',
+        img_prefix=f'{data_root}/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/apt36k/ViTPose_small_apt36k_256x192.py b/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/apt36k/ViTPose_small_apt36k_256x192.py
new file mode 100644
index 0000000..691d373
--- /dev/null
+++ b/configs/animal/2d_kpt_sview_rgb_img/topdown_heatmap/apt36k/ViTPose_small_apt36k_256x192.py
@@ -0,0 +1,157 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/ap10k.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+log_config = dict(
+    interval=1,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=17,
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=384,
+        depth=12,
+        num_heads=12,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=384,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=True,
+    det_bbox_thr=0.0,
+    bbox_file='',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/apt36k'
+data = dict(
+    samples_per_gpu=32,
+    workers_per_gpu=4,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/train_annotations_1.json',
+        img_prefix=f'{data_root}/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/val_annotations_1.json',
+        img_prefix=f'{data_root}/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='AnimalAP10KDataset',
+        ann_file=f'{data_root}/annotations/val_annotations_1.json',
+        img_prefix=f'{data_root}/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
\ No newline at end of file
diff --git a/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/aic/ViTPose_small_aic_256x192.py b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/aic/ViTPose_small_aic_256x192.py
new file mode 100644
index 0000000..af66009
--- /dev/null
+++ b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/aic/ViTPose_small_aic_256x192.py
@@ -0,0 +1,151 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/aic.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+channel_cfg = dict(
+    num_output_channels=14,
+    dataset_joints=14,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13],
+    ],
+    inference_channel=[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=384,
+        depth=12,
+        num_heads=12,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=384,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=True,
+    det_bbox_thr=0.0,
+    bbox_file='data/coco/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/aic'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=2,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='TopDownAicDataset',
+        ann_file=f'{data_root}/annotations/aic_train.json',
+        img_prefix=f'{data_root}/ai_challenger_keypoint_train_20170902/'
+        'keypoint_train_images_20170902/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='TopDownAicDataset',
+        ann_file=f'{data_root}/annotations/aic_val.json',
+        img_prefix=f'{data_root}/ai_challenger_keypoint_validation_20170911/'
+        'keypoint_validation_images_20170911/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='TopDownAicDataset',
+        ann_file=f'{data_root}/annotations/aic_val.json',
+        img_prefix=f'{data_root}/ai_challenger_keypoint_validation_20170911/'
+        'keypoint_validation_images_20170911/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}))
diff --git a/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_small_coco_256x192.py b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_small_coco_256x192.py
new file mode 100644
index 0000000..bec19fa
--- /dev/null
+++ b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_small_coco_256x192.py
@@ -0,0 +1,170 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/coco.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(type='AdamW', lr=5e-4, betas=(0.9, 0.999), weight_decay=0.1,
+                 constructor='LayerDecayOptimizerConstructor', 
+                 paramwise_cfg=dict(
+                                    num_layers=12, 
+                                    layer_decay_rate=0.9,
+                                    custom_keys={
+                                            'bias': dict(decay_multi=0.),
+                                            'pos_embed': dict(decay_mult=0.),
+                                            'relative_position_bias_table': dict(decay_mult=0.),
+                                            'norm': dict(decay_mult=0.)
+                                            }
+                                    )
+                )
+
+optimizer_config = dict(grad_clip=dict(max_norm=1., norm_type=2))
+
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+target_type = 'GaussianHeatmap'
+channel_cfg = dict(
+    num_output_channels=17,
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=384,
+        depth=12,
+        num_heads=12,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=384,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=False,
+        target_type=target_type,
+        modulate_kernel=11,
+        use_udp=True))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=False,
+    det_bbox_thr=0.0,
+    bbox_file='data/coco/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine', use_udp=True),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='TopDownGenerateTarget',
+        sigma=2,
+        encoding='UDP',
+        target_type=target_type),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine', use_udp=True),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/coco'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=4,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='TopDownCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
+        img_prefix=f'{data_root}/train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='TopDownCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='TopDownCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
+
diff --git a/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_small_simple_coco_256x192.py b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_small_simple_coco_256x192.py
new file mode 100644
index 0000000..bec19fa
--- /dev/null
+++ b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_small_simple_coco_256x192.py
@@ -0,0 +1,170 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/coco.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(type='AdamW', lr=5e-4, betas=(0.9, 0.999), weight_decay=0.1,
+                 constructor='LayerDecayOptimizerConstructor', 
+                 paramwise_cfg=dict(
+                                    num_layers=12, 
+                                    layer_decay_rate=0.9,
+                                    custom_keys={
+                                            'bias': dict(decay_multi=0.),
+                                            'pos_embed': dict(decay_mult=0.),
+                                            'relative_position_bias_table': dict(decay_mult=0.),
+                                            'norm': dict(decay_mult=0.)
+                                            }
+                                    )
+                )
+
+optimizer_config = dict(grad_clip=dict(max_norm=1., norm_type=2))
+
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+target_type = 'GaussianHeatmap'
+channel_cfg = dict(
+    num_output_channels=17,
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=384,
+        depth=12,
+        num_heads=12,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=384,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=False,
+        target_type=target_type,
+        modulate_kernel=11,
+        use_udp=True))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=False,
+    det_bbox_thr=0.0,
+    bbox_file='data/coco/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine', use_udp=True),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='TopDownGenerateTarget',
+        sigma=2,
+        encoding='UDP',
+        target_type=target_type),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine', use_udp=True),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/coco'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=4,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='TopDownCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
+        img_prefix=f'{data_root}/train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='TopDownCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='TopDownCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
+
diff --git a/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/vitPose+_base_coco+aic+mpii+ap10k+apt36k+wholebody_256x192_udp.py b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/vitPose+_base_coco+aic+mpii+ap10k+apt36k+wholebody_256x192_udp.py
index 01677a4..391ab15 100644
--- a/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/vitPose+_base_coco+aic+mpii+ap10k+apt36k+wholebody_256x192_udp.py
+++ b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/vitPose+_base_coco+aic+mpii+ap10k+apt36k+wholebody_256x192_udp.py
@@ -126,15 +126,6 @@ model = dict(
             extra=dict(final_conv_kernel=1, ),
             out_channels=mpii_channel_cfg['num_output_channels'],
             loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
-        dict(
-            type='TopdownHeatmapSimpleHead',
-            in_channels=768,
-            num_deconv_layers=2,
-            num_deconv_filters=(256, 256),
-            num_deconv_kernels=(4, 4),
-            extra=dict(final_conv_kernel=1, ),
-            out_channels=crowdpose_channel_cfg['num_output_channels'],
-            loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
         dict(
             type='TopdownHeatmapSimpleHead',
             in_channels=768,
diff --git a/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/vitPose+_huge_coco+aic+mpii+ap10k+apt36k+wholebody_256x192_udp.py b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/vitPose+_huge_coco+aic+mpii+ap10k+apt36k+wholebody_256x192_udp.py
index 2b9dbee..612aaf0 100644
--- a/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/vitPose+_huge_coco+aic+mpii+ap10k+apt36k+wholebody_256x192_udp.py
+++ b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/vitPose+_huge_coco+aic+mpii+ap10k+apt36k+wholebody_256x192_udp.py
@@ -126,15 +126,6 @@ model = dict(
             extra=dict(final_conv_kernel=1, ),
             out_channels=mpii_channel_cfg['num_output_channels'],
             loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
-        dict(
-            type='TopdownHeatmapSimpleHead',
-            in_channels=1280,
-            num_deconv_layers=2,
-            num_deconv_filters=(256, 256),
-            num_deconv_kernels=(4, 4),
-            extra=dict(final_conv_kernel=1, ),
-            out_channels=crowdpose_channel_cfg['num_output_channels'],
-            loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
         dict(
             type='TopdownHeatmapSimpleHead',
             in_channels=1280,
diff --git a/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/vitPose+_large_coco+aic+mpii+ap10k+apt36k+wholebody_256x192_udp.py b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/vitPose+_large_coco+aic+mpii+ap10k+apt36k+wholebody_256x192_udp.py
index 98b3d4d..0936de4 100644
--- a/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/vitPose+_large_coco+aic+mpii+ap10k+apt36k+wholebody_256x192_udp.py
+++ b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/vitPose+_large_coco+aic+mpii+ap10k+apt36k+wholebody_256x192_udp.py
@@ -126,15 +126,6 @@ model = dict(
             extra=dict(final_conv_kernel=1, ),
             out_channels=mpii_channel_cfg['num_output_channels'],
             loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
-        dict(
-            type='TopdownHeatmapSimpleHead',
-            in_channels=1024,
-            num_deconv_layers=2,
-            num_deconv_filters=(256, 256),
-            num_deconv_kernels=(4, 4),
-            extra=dict(final_conv_kernel=1, ),
-            out_channels=crowdpose_channel_cfg['num_output_channels'],
-            loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
         dict(
             type='TopdownHeatmapSimpleHead',
             in_channels=1024,
diff --git a/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/vitPose+_small_coco+aic+mpii+ap10k+apt36k+wholebody_256x192_udp.py b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/vitPose+_small_coco+aic+mpii+ap10k+apt36k+wholebody_256x192_udp.py
new file mode 100644
index 0000000..34823a3
--- /dev/null
+++ b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/vitPose+_small_coco+aic+mpii+ap10k+apt36k+wholebody_256x192_udp.py
@@ -0,0 +1,491 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/coco.py',
+    '../../../../_base_/datasets/aic_info.py',
+    '../../../../_base_/datasets/mpii_info.py',
+    '../../../../_base_/datasets/ap10k_info.py',
+    '../../../../_base_/datasets/coco_wholebody_info.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(type='AdamW', lr=1e-3, betas=(0.9, 0.999), weight_decay=0.1,
+                 constructor='LayerDecayOptimizerConstructor', 
+                 paramwise_cfg=dict(
+                                    num_layers=12, 
+                                    layer_decay_rate=0.9,
+                                    custom_keys={
+                                            'bias': dict(decay_multi=0.),
+                                            'pos_embed': dict(decay_mult=0.),
+                                            'relative_position_bias_table': dict(decay_mult=0.),
+                                            'norm': dict(decay_mult=0.)
+                                            }
+                                    )
+                )
+
+optimizer_config = dict(grad_clip=dict(max_norm=1., norm_type=2))
+
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+target_type = 'GaussianHeatmap'
+channel_cfg = dict(
+    num_output_channels=17,
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+aic_channel_cfg = dict(
+    num_output_channels=14,
+    dataset_joints=14,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13],
+    ],
+    inference_channel=[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13])
+mpii_channel_cfg = dict(
+    num_output_channels=16,
+    dataset_joints=16,
+    dataset_channel=list(range(16)),
+    inference_channel=list(range(16)))
+crowdpose_channel_cfg = dict(
+    num_output_channels=14,
+    dataset_joints=14,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13],
+    ],
+    inference_channel=[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13])
+ap10k_channel_cfg = dict(
+    num_output_channels=17,
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+cocowholebody_channel_cfg = dict(
+    num_output_channels=133,
+    dataset_joints=133,
+    dataset_channel=[
+        list(range(133)),
+    ],
+    inference_channel=list(range(133)))
+
+
+# model settings
+model = dict(
+    type='TopDownMoE',
+    pretrained=None,
+    backbone=dict(
+        type='ViTMoE',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=384,
+        depth=12,
+        num_heads=12,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+        num_expert=6,
+        part_features=192
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=384,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    associate_keypoint_head=[
+        dict(
+            type='TopdownHeatmapSimpleHead',
+            in_channels=384,
+            num_deconv_layers=2,
+            num_deconv_filters=(256, 256),
+            num_deconv_kernels=(4, 4),
+            extra=dict(final_conv_kernel=1, ),
+            out_channels=aic_channel_cfg['num_output_channels'],
+            loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+        dict(
+            type='TopdownHeatmapSimpleHead',
+            in_channels=384,
+            num_deconv_layers=2,
+            num_deconv_filters=(256, 256),
+            num_deconv_kernels=(4, 4),
+            extra=dict(final_conv_kernel=1, ),
+            out_channels=mpii_channel_cfg['num_output_channels'],
+            loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+        dict(
+            type='TopdownHeatmapSimpleHead',
+            in_channels=384,
+            num_deconv_layers=2,
+            num_deconv_filters=(256, 256),
+            num_deconv_kernels=(4, 4),
+            extra=dict(final_conv_kernel=1, ),
+            out_channels=ap10k_channel_cfg['num_output_channels'],
+            loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+        dict(
+            type='TopdownHeatmapSimpleHead',
+            in_channels=384,
+            num_deconv_layers=2,
+            num_deconv_filters=(256, 256),
+            num_deconv_kernels=(4, 4),
+            extra=dict(final_conv_kernel=1, ),
+            out_channels=ap10k_channel_cfg['num_output_channels'],
+            loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+        dict(
+            type='TopdownHeatmapSimpleHead',
+            in_channels=384,
+            num_deconv_layers=2,
+            num_deconv_filters=(256, 256),
+            num_deconv_kernels=(4, 4),
+            extra=dict(final_conv_kernel=1, ),
+            out_channels=cocowholebody_channel_cfg['num_output_channels'],
+            loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+        ],
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=False,
+        target_type=target_type,
+        modulate_kernel=11,
+        use_udp=True))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=False,
+    det_bbox_thr=0.0,
+    bbox_file='data/coco/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+    max_num_joints=133,
+    dataset_idx=0,
+)
+
+aic_data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=aic_channel_cfg['num_output_channels'],
+    num_joints=aic_channel_cfg['dataset_joints'],
+    dataset_channel=aic_channel_cfg['dataset_channel'],
+    inference_channel=aic_channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=True,
+    det_bbox_thr=0.0,
+    bbox_file='data/coco/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+    max_num_joints=133,
+    dataset_idx=1,
+)
+
+mpii_data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=mpii_channel_cfg['num_output_channels'],
+    num_joints=mpii_channel_cfg['dataset_joints'],
+    dataset_channel=mpii_channel_cfg['dataset_channel'],
+    inference_channel=mpii_channel_cfg['inference_channel'],
+    max_num_joints=133,
+    dataset_idx=2,
+    use_gt_bbox=True,
+    bbox_file=None,
+)
+
+ap10k_data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=True,
+    det_bbox_thr=0.0,
+    bbox_file='',
+    max_num_joints=133,
+    dataset_idx=3,
+)
+
+ap36k_data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=True,
+    det_bbox_thr=0.0,
+    bbox_file='',
+    max_num_joints=133,
+    dataset_idx=4,
+)
+
+cocowholebody_data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=cocowholebody_channel_cfg['num_output_channels'],
+    num_joints=cocowholebody_channel_cfg['dataset_joints'],
+    dataset_channel=cocowholebody_channel_cfg['dataset_channel'],
+    inference_channel=cocowholebody_channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=False,
+    det_bbox_thr=0.0,
+    bbox_file='data/coco/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+    dataset_idx=5,
+    max_num_joints=133,
+)
+
+cocowholebody_train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs', 'dataset_idx'
+        ]),
+]
+
+ap10k_train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs', 'dataset_idx'
+        ]),
+]
+
+aic_train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs', 'dataset_idx'
+        ]),
+]
+
+mpii_train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine', use_udp=True),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='TopDownGenerateTarget',
+        sigma=2,
+        encoding='UDP',
+        target_type=target_type),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'flip_pairs', 'dataset_idx'
+        ]),
+]
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine', use_udp=True),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='TopDownGenerateTarget',
+        sigma=2,
+        encoding='UDP',
+        target_type=target_type),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs', 'dataset_idx'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine', use_udp=True),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs', 'dataset_idx'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/coco'
+aic_data_root = 'data/aic'
+mpii_data_root = 'data/mpii'
+ap10k_data_root = 'data/ap10k'
+ap36k_data_root = 'data/ap36k'
+
+data = dict(
+    samples_per_gpu=128,
+    workers_per_gpu=8,
+    val_dataloader=dict(samples_per_gpu=64),
+    test_dataloader=dict(samples_per_gpu=64),
+    train=[
+        dict(
+            type='TopDownCocoDataset',
+            ann_file=f'{data_root}/annotations/person_keypoints_train2017.json',
+            img_prefix=f'{data_root}/train2017/',
+            data_cfg=data_cfg,
+            pipeline=train_pipeline,
+            dataset_info={{_base_.dataset_info}}),
+        dict(
+            type='TopDownAicDataset',
+            ann_file=f'{aic_data_root}/annotations/person_keypoints_train2017.json',
+            img_prefix=f'{aic_data_root}/ai_challenger_keypoint_train_20170909/'
+            'keypoint_train_images_20170902/',
+            data_cfg=aic_data_cfg,
+            pipeline=aic_train_pipeline,
+            dataset_info={{_base_.aic_info}}),
+        dict(
+            type='TopDownMpiiDataset',
+            ann_file=f'{mpii_data_root}/annotations/mpii_train.json',
+            img_prefix=f'{mpii_data_root}/images/',
+            data_cfg=mpii_data_cfg,
+            pipeline=mpii_train_pipeline,
+            dataset_info={{_base_.mpii_info}}),
+        dict(
+            type='AnimalAP10KDataset',
+            ann_file=f'{ap10k_data_root}/annotations/ap10k-train-split1.json',
+            img_prefix=f'{ap10k_data_root}/data/',
+            data_cfg=ap10k_data_cfg,
+            pipeline=ap10k_train_pipeline,
+            dataset_info={{_base_.ap10k_info}}),
+        dict(
+            type='AnimalAP10KDataset',
+            ann_file=f'{ap36k_data_root}/annotations/train_annotations_1.json',
+            img_prefix=f'{ap36k_data_root}/',
+            data_cfg=ap36k_data_cfg,
+            pipeline=ap10k_train_pipeline,
+            dataset_info={{_base_.ap10k_info}}),
+        dict(
+            type='TopDownCocoWholeBodyDataset',
+            ann_file=f'{data_root}/annotations/coco_wholebody_train_v1.0.json',
+            img_prefix=f'{data_root}/train2017/',
+            data_cfg=cocowholebody_data_cfg,
+            pipeline=cocowholebody_train_pipeline,
+            dataset_info={{_base_.cocowholebody_info}}),
+        ],
+    val=dict(
+        type='TopDownCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='TopDownCocoDataset',
+        ann_file=f'{data_root}/annotations/person_keypoints_val2017.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
+
diff --git a/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/mpii/ViTPose_small_mpii_256x192.py b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/mpii/ViTPose_small_mpii_256x192.py
new file mode 100644
index 0000000..f80f522
--- /dev/null
+++ b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/mpii/ViTPose_small_mpii_256x192.py
@@ -0,0 +1,146 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/mpii.py'
+]
+evaluation = dict(interval=10, metric='PCKh', save_best='PCKh')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+log_config = dict(
+    interval=50, hooks=[
+        dict(type='TextLoggerHook'),
+    ])
+
+target_type = 'GaussianHeatmap'
+channel_cfg = dict(
+    num_output_channels=16,
+    dataset_joints=16,
+    dataset_channel=list(range(16)),
+    inference_channel=list(range(16)))
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=384,
+        depth=12,
+        num_heads=12,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=384,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=False,
+        target_type=target_type,
+        modulate_kernel=11,
+        use_udp=True))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    use_gt_bbox=True,
+    bbox_file=None,
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine', use_udp=True),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='TopDownGenerateTarget',
+        sigma=2,
+        encoding='UDP',
+        target_type=target_type),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine', use_udp=True),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=['image_file', 'center', 'scale', 'rotation', 'flip_pairs']),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/mpii'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=2,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='TopDownMpiiDataset',
+        ann_file=f'{data_root}/annotations/mpii_train.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='TopDownMpiiDataset',
+        ann_file=f'{data_root}/annotations/mpii_val.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='TopDownMpiiDataset',
+        ann_file=f'{data_root}/annotations/mpii_val.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_small_ochuman_256x192.py b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_small_ochuman_256x192.py
new file mode 100644
index 0000000..58bd1ca
--- /dev/null
+++ b/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/ochuman/ViTPose_small_ochuman_256x192.py
@@ -0,0 +1,153 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/ochuman.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+channel_cfg = dict(
+    num_output_channels=17,
+    dataset_joints=17,
+    dataset_channel=[
+        [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=384,
+        depth=12,
+        num_heads=12,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=384,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=True,
+    det_bbox_thr=0.0,
+    bbox_file='data/coco/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/ochuman'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=2,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='TopDownCocoDataset',
+        ann_file='data/coco/annotations/person_keypoints_train2017.json',
+        img_prefix='data/coco//train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='TopDownOCHumanDataset',
+        ann_file=f'{data_root}/annotations/'
+        'ochuman_coco_format_val_range_0.00_1.00.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='TopDownOCHumanDataset',
+        ann_file=f'{data_root}/annotations/'
+        'ochuman_coco_format_test_range_0.00_1.00.json',
+        img_prefix=f'{data_root}/images/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/configs/hand/2d_kpt_sview_rgb_img/topdown_heatmap/interhand2d/ViTPose_base_interhand2d_all_256x192.py b/configs/hand/2d_kpt_sview_rgb_img/topdown_heatmap/interhand2d/ViTPose_base_interhand2d_all_256x192.py
new file mode 100644
index 0000000..275b3a3
--- /dev/null
+++ b/configs/hand/2d_kpt_sview_rgb_img/topdown_heatmap/interhand2d/ViTPose_base_interhand2d_all_256x192.py
@@ -0,0 +1,162 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/interhand2d.py'
+]
+checkpoint_config = dict(interval=5)
+evaluation = dict(interval=5, metric=['PCK', 'AUC', 'EPE'], save_best='AUC')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[40, 50])
+total_epochs = 60
+log_config = dict(
+    interval=20,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=21,
+    dataset_joints=21,
+    dataset_channel=[
+        [
+            0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18,
+            19, 20
+        ],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19,
+        20
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=768,
+        depth=12,
+        num_heads=12,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=768,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'])
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=90, scale_factor=0.3),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=['image_file', 'center', 'scale', 'rotation', 'flip_pairs']),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/interhand2.6m'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=2,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='InterHand2DDataset',
+        ann_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_train_data.json',
+        camera_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_train_camera.json',
+        joint_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_train_joint_3d.json',
+        img_prefix=f'{data_root}/images/train/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='InterHand2DDataset',
+        ann_file=f'{data_root}/annotations/machine_annot/'
+        'InterHand2.6M_val_data.json',
+        camera_file=f'{data_root}/annotations/machine_annot/'
+        'InterHand2.6M_val_camera.json',
+        joint_file=f'{data_root}/annotations/machine_annot/'
+        'InterHand2.6M_val_joint_3d.json',
+        img_prefix=f'{data_root}/images/val/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='InterHand2DDataset',
+        ann_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_test_data.json',
+        camera_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_test_camera.json',
+        joint_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_test_joint_3d.json',
+        img_prefix=f'{data_root}/images/test/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/configs/hand/2d_kpt_sview_rgb_img/topdown_heatmap/interhand2d/ViTPose_huge_interhand2d_all_256x192.py b/configs/hand/2d_kpt_sview_rgb_img/topdown_heatmap/interhand2d/ViTPose_huge_interhand2d_all_256x192.py
new file mode 100644
index 0000000..2af0f77
--- /dev/null
+++ b/configs/hand/2d_kpt_sview_rgb_img/topdown_heatmap/interhand2d/ViTPose_huge_interhand2d_all_256x192.py
@@ -0,0 +1,162 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/interhand2d.py'
+]
+checkpoint_config = dict(interval=5)
+evaluation = dict(interval=5, metric=['PCK', 'AUC', 'EPE'], save_best='AUC')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[40, 50])
+total_epochs = 60
+log_config = dict(
+    interval=20,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=21,
+    dataset_joints=21,
+    dataset_channel=[
+        [
+            0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18,
+            19, 20
+        ],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19,
+        20
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=1280,
+        depth=32,
+        num_heads=16,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=1280,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'])
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=90, scale_factor=0.3),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=['image_file', 'center', 'scale', 'rotation', 'flip_pairs']),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/interhand2.6m'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=2,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='InterHand2DDataset',
+        ann_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_train_data.json',
+        camera_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_train_camera.json',
+        joint_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_train_joint_3d.json',
+        img_prefix=f'{data_root}/images/train/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='InterHand2DDataset',
+        ann_file=f'{data_root}/annotations/machine_annot/'
+        'InterHand2.6M_val_data.json',
+        camera_file=f'{data_root}/annotations/machine_annot/'
+        'InterHand2.6M_val_camera.json',
+        joint_file=f'{data_root}/annotations/machine_annot/'
+        'InterHand2.6M_val_joint_3d.json',
+        img_prefix=f'{data_root}/images/val/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='InterHand2DDataset',
+        ann_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_test_data.json',
+        camera_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_test_camera.json',
+        joint_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_test_joint_3d.json',
+        img_prefix=f'{data_root}/images/test/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/configs/hand/2d_kpt_sview_rgb_img/topdown_heatmap/interhand2d/ViTPose_large_interhand2d_all_256x192.py b/configs/hand/2d_kpt_sview_rgb_img/topdown_heatmap/interhand2d/ViTPose_large_interhand2d_all_256x192.py
new file mode 100644
index 0000000..72c33a7
--- /dev/null
+++ b/configs/hand/2d_kpt_sview_rgb_img/topdown_heatmap/interhand2d/ViTPose_large_interhand2d_all_256x192.py
@@ -0,0 +1,162 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/interhand2d.py'
+]
+checkpoint_config = dict(interval=5)
+evaluation = dict(interval=5, metric=['PCK', 'AUC', 'EPE'], save_best='AUC')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[40, 50])
+total_epochs = 60
+log_config = dict(
+    interval=20,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=21,
+    dataset_joints=21,
+    dataset_channel=[
+        [
+            0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18,
+            19, 20
+        ],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19,
+        20
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=1024,
+        depth=24,
+        num_heads=16,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=1024,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'])
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=90, scale_factor=0.3),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=['image_file', 'center', 'scale', 'rotation', 'flip_pairs']),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/interhand2.6m'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=2,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='InterHand2DDataset',
+        ann_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_train_data.json',
+        camera_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_train_camera.json',
+        joint_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_train_joint_3d.json',
+        img_prefix=f'{data_root}/images/train/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='InterHand2DDataset',
+        ann_file=f'{data_root}/annotations/machine_annot/'
+        'InterHand2.6M_val_data.json',
+        camera_file=f'{data_root}/annotations/machine_annot/'
+        'InterHand2.6M_val_camera.json',
+        joint_file=f'{data_root}/annotations/machine_annot/'
+        'InterHand2.6M_val_joint_3d.json',
+        img_prefix=f'{data_root}/images/val/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='InterHand2DDataset',
+        ann_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_test_data.json',
+        camera_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_test_camera.json',
+        joint_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_test_joint_3d.json',
+        img_prefix=f'{data_root}/images/test/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/configs/hand/2d_kpt_sview_rgb_img/topdown_heatmap/interhand2d/ViTPose_small_interhand2d_all_256x192.py b/configs/hand/2d_kpt_sview_rgb_img/topdown_heatmap/interhand2d/ViTPose_small_interhand2d_all_256x192.py
new file mode 100644
index 0000000..d344dca
--- /dev/null
+++ b/configs/hand/2d_kpt_sview_rgb_img/topdown_heatmap/interhand2d/ViTPose_small_interhand2d_all_256x192.py
@@ -0,0 +1,162 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/interhand2d.py'
+]
+checkpoint_config = dict(interval=5)
+evaluation = dict(interval=5, metric=['PCK', 'AUC', 'EPE'], save_best='AUC')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[40, 50])
+total_epochs = 60
+log_config = dict(
+    interval=20,
+    hooks=[
+        dict(type='TextLoggerHook'),
+        # dict(type='TensorboardLoggerHook')
+    ])
+
+channel_cfg = dict(
+    num_output_channels=21,
+    dataset_joints=21,
+    dataset_channel=[
+        [
+            0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18,
+            19, 20
+        ],
+    ],
+    inference_channel=[
+        0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19,
+        20
+    ])
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=384,
+        depth=12,
+        num_heads=12,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=384,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'])
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=90, scale_factor=0.3),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=['image_file', 'center', 'scale', 'rotation', 'flip_pairs']),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/interhand2.6m'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=2,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='InterHand2DDataset',
+        ann_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_train_data.json',
+        camera_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_train_camera.json',
+        joint_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_train_joint_3d.json',
+        img_prefix=f'{data_root}/images/train/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='InterHand2DDataset',
+        ann_file=f'{data_root}/annotations/machine_annot/'
+        'InterHand2.6M_val_data.json',
+        camera_file=f'{data_root}/annotations/machine_annot/'
+        'InterHand2.6M_val_camera.json',
+        joint_file=f'{data_root}/annotations/machine_annot/'
+        'InterHand2.6M_val_joint_3d.json',
+        img_prefix=f'{data_root}/images/val/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='InterHand2DDataset',
+        ann_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_test_data.json',
+        camera_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_test_camera.json',
+        joint_file=f'{data_root}/annotations/all/'
+        'InterHand2.6M_test_joint_3d.json',
+        img_prefix=f'{data_root}/images/test/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/configs/wholebody/2d_kpt_sview_rgb_img/topdown_heatmap/coco-wholebody/ViTPose_base_wholebody_256x192.py b/configs/wholebody/2d_kpt_sview_rgb_img/topdown_heatmap/coco-wholebody/ViTPose_base_wholebody_256x192.py
new file mode 100644
index 0000000..02db322
--- /dev/null
+++ b/configs/wholebody/2d_kpt_sview_rgb_img/topdown_heatmap/coco-wholebody/ViTPose_base_wholebody_256x192.py
@@ -0,0 +1,149 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/coco_wholebody.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+channel_cfg = dict(
+    num_output_channels=133,
+    dataset_joints=133,
+    dataset_channel=[
+        list(range(133)),
+    ],
+    inference_channel=list(range(133)))
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=768,
+        depth=12,
+        num_heads=12,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=768,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=False,
+    det_bbox_thr=0.0,
+    bbox_file='data/coco/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/coco'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=2,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='TopDownCocoWholeBodyDataset',
+        ann_file=f'{data_root}/annotations/coco_wholebody_train_v1.0.json',
+        img_prefix=f'{data_root}/train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='TopDownCocoWholeBodyDataset',
+        ann_file=f'{data_root}/annotations/coco_wholebody_val_v1.0.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='TopDownCocoWholeBodyDataset',
+        ann_file=f'{data_root}/annotations/coco_wholebody_val_v1.0.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/configs/wholebody/2d_kpt_sview_rgb_img/topdown_heatmap/coco-wholebody/ViTPose_huge_wholebody_256x192.py b/configs/wholebody/2d_kpt_sview_rgb_img/topdown_heatmap/coco-wholebody/ViTPose_huge_wholebody_256x192.py
new file mode 100644
index 0000000..ccd8fd2
--- /dev/null
+++ b/configs/wholebody/2d_kpt_sview_rgb_img/topdown_heatmap/coco-wholebody/ViTPose_huge_wholebody_256x192.py
@@ -0,0 +1,149 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/coco_wholebody.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+channel_cfg = dict(
+    num_output_channels=133,
+    dataset_joints=133,
+    dataset_channel=[
+        list(range(133)),
+    ],
+    inference_channel=list(range(133)))
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=1280,
+        depth=32,
+        num_heads=16,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=1280,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=False,
+    det_bbox_thr=0.0,
+    bbox_file='data/coco/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/coco'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=2,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='TopDownCocoWholeBodyDataset',
+        ann_file=f'{data_root}/annotations/coco_wholebody_train_v1.0.json',
+        img_prefix=f'{data_root}/train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='TopDownCocoWholeBodyDataset',
+        ann_file=f'{data_root}/annotations/coco_wholebody_val_v1.0.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='TopDownCocoWholeBodyDataset',
+        ann_file=f'{data_root}/annotations/coco_wholebody_val_v1.0.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/configs/wholebody/2d_kpt_sview_rgb_img/topdown_heatmap/coco-wholebody/ViTPose_large_wholebody_256x192.py b/configs/wholebody/2d_kpt_sview_rgb_img/topdown_heatmap/coco-wholebody/ViTPose_large_wholebody_256x192.py
new file mode 100644
index 0000000..df96867
--- /dev/null
+++ b/configs/wholebody/2d_kpt_sview_rgb_img/topdown_heatmap/coco-wholebody/ViTPose_large_wholebody_256x192.py
@@ -0,0 +1,149 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/coco_wholebody.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+channel_cfg = dict(
+    num_output_channels=133,
+    dataset_joints=133,
+    dataset_channel=[
+        list(range(133)),
+    ],
+    inference_channel=list(range(133)))
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=1024,
+        depth=24,
+        num_heads=16,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=1024,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=False,
+    det_bbox_thr=0.0,
+    bbox_file='data/coco/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/coco'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=2,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='TopDownCocoWholeBodyDataset',
+        ann_file=f'{data_root}/annotations/coco_wholebody_train_v1.0.json',
+        img_prefix=f'{data_root}/train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='TopDownCocoWholeBodyDataset',
+        ann_file=f'{data_root}/annotations/coco_wholebody_val_v1.0.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='TopDownCocoWholeBodyDataset',
+        ann_file=f'{data_root}/annotations/coco_wholebody_val_v1.0.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/configs/wholebody/2d_kpt_sview_rgb_img/topdown_heatmap/coco-wholebody/ViTPose_small_wholebody_256x192.py b/configs/wholebody/2d_kpt_sview_rgb_img/topdown_heatmap/coco-wholebody/ViTPose_small_wholebody_256x192.py
new file mode 100644
index 0000000..d1d4b05
--- /dev/null
+++ b/configs/wholebody/2d_kpt_sview_rgb_img/topdown_heatmap/coco-wholebody/ViTPose_small_wholebody_256x192.py
@@ -0,0 +1,149 @@
+_base_ = [
+    '../../../../_base_/default_runtime.py',
+    '../../../../_base_/datasets/coco_wholebody.py'
+]
+evaluation = dict(interval=10, metric='mAP', save_best='AP')
+
+optimizer = dict(
+    type='Adam',
+    lr=5e-4,
+)
+optimizer_config = dict(grad_clip=None)
+# learning policy
+lr_config = dict(
+    policy='step',
+    warmup='linear',
+    warmup_iters=500,
+    warmup_ratio=0.001,
+    step=[170, 200])
+total_epochs = 210
+channel_cfg = dict(
+    num_output_channels=133,
+    dataset_joints=133,
+    dataset_channel=[
+        list(range(133)),
+    ],
+    inference_channel=list(range(133)))
+
+# model settings
+model = dict(
+    type='TopDown',
+    pretrained=None,
+    backbone=dict(
+        type='ViT',
+        img_size=(256, 192),
+        patch_size=16,
+        embed_dim=384,
+        depth=12,
+        num_heads=12,
+        ratio=1,
+        use_checkpoint=False,
+        mlp_ratio=4,
+        qkv_bias=True,
+        drop_path_rate=0.3,
+    ),
+    keypoint_head=dict(
+        type='TopdownHeatmapSimpleHead',
+        in_channels=384,
+        num_deconv_layers=2,
+        num_deconv_filters=(256, 256),
+        num_deconv_kernels=(4, 4),
+        extra=dict(final_conv_kernel=1, ),
+        out_channels=channel_cfg['num_output_channels'],
+        loss_keypoint=dict(type='JointsMSELoss', use_target_weight=True)),
+    train_cfg=dict(),
+    test_cfg=dict(
+        flip_test=True,
+        post_process='default',
+        shift_heatmap=True,
+        modulate_kernel=11))
+
+data_cfg = dict(
+    image_size=[192, 256],
+    heatmap_size=[48, 64],
+    num_output_channels=channel_cfg['num_output_channels'],
+    num_joints=channel_cfg['dataset_joints'],
+    dataset_channel=channel_cfg['dataset_channel'],
+    inference_channel=channel_cfg['inference_channel'],
+    soft_nms=False,
+    nms_thr=1.0,
+    oks_thr=0.9,
+    vis_thr=0.2,
+    use_gt_bbox=False,
+    det_bbox_thr=0.0,
+    bbox_file='data/coco/person_detection_results/'
+    'COCO_val2017_detections_AP_H_56_person.json',
+)
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownRandomFlip', flip_prob=0.5),
+    dict(
+        type='TopDownHalfBodyTransform',
+        num_joints_half_body=8,
+        prob_half_body=0.3),
+    dict(
+        type='TopDownGetRandomScaleRotation', rot_factor=40, scale_factor=0.5),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(type='TopDownGenerateTarget', sigma=2),
+    dict(
+        type='Collect',
+        keys=['img', 'target', 'target_weight'],
+        meta_keys=[
+            'image_file', 'joints_3d', 'joints_3d_visible', 'center', 'scale',
+            'rotation', 'bbox_score', 'flip_pairs'
+        ]),
+]
+
+val_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='TopDownAffine'),
+    dict(type='ToTensor'),
+    dict(
+        type='NormalizeTensor',
+        mean=[0.485, 0.456, 0.406],
+        std=[0.229, 0.224, 0.225]),
+    dict(
+        type='Collect',
+        keys=['img'],
+        meta_keys=[
+            'image_file', 'center', 'scale', 'rotation', 'bbox_score',
+            'flip_pairs'
+        ]),
+]
+
+test_pipeline = val_pipeline
+
+data_root = 'data/coco'
+data = dict(
+    samples_per_gpu=64,
+    workers_per_gpu=2,
+    val_dataloader=dict(samples_per_gpu=32),
+    test_dataloader=dict(samples_per_gpu=32),
+    train=dict(
+        type='TopDownCocoWholeBodyDataset',
+        ann_file=f'{data_root}/annotations/coco_wholebody_train_v1.0.json',
+        img_prefix=f'{data_root}/train2017/',
+        data_cfg=data_cfg,
+        pipeline=train_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    val=dict(
+        type='TopDownCocoWholeBodyDataset',
+        ann_file=f'{data_root}/annotations/coco_wholebody_val_v1.0.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=val_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+    test=dict(
+        type='TopDownCocoWholeBodyDataset',
+        ann_file=f'{data_root}/annotations/coco_wholebody_val_v1.0.json',
+        img_prefix=f'{data_root}/val2017/',
+        data_cfg=data_cfg,
+        pipeline=test_pipeline,
+        dataset_info={{_base_.dataset_info}}),
+)
diff --git a/figures/Throughput.png b/figures/Throughput.png
index 3f26b4c..b13edca 100644
Binary files a/figures/Throughput.png and b/figures/Throughput.png differ
diff --git a/mmpose/models/backbones/vit_moe.py b/mmpose/models/backbones/vit_moe.py
index 2daa270..880a58f 100644
--- a/mmpose/models/backbones/vit_moe.py
+++ b/mmpose/models/backbones/vit_moe.py
@@ -171,7 +171,8 @@ class Block(nn.Module):
         self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()
         self.norm2 = norm_layer(dim)
         mlp_hidden_dim = int(dim * mlp_ratio)
-        self.mlp = MoEMlp(num_expert=num_expert, in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop, mode=mode)
+        self.mlp = MoEMlp(num_expert=num_expert, in_features=dim, hidden_features=mlp_hidden_dim, 
+                            act_layer=act_layer, drop=drop, part_features=part_features)
 
     def forward(self, x, indices=None):
 
diff --git a/tools/model_split.py b/tools/model_split.py
new file mode 100644
index 0000000..3cfe13f
--- /dev/null
+++ b/tools/model_split.py
@@ -0,0 +1,93 @@
+import torch
+import os
+import argparse
+import copy
+
+def parse_args():
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--source', type=str)
+    parser.add_argument('--target', type=str, default=None)
+    args = parser.parse_args()
+    return args
+
+def main():
+
+    args = parse_args()
+
+    if args.target is None:
+        args.target = '/'.join(args.source.split('/')[:-1])
+
+    ckpt = torch.load(args.source, map_location='cpu')
+    
+    experts = dict()
+
+    new_ckpt = copy.deepcopy(ckpt)
+
+    state_dict = new_ckpt['state_dict']
+
+    for key, value in state_dict.items():
+        if 'mlp.experts' in key:
+            experts[key] = value
+
+    keys = ckpt['state_dict'].keys()
+
+    target_expert = 0
+    new_ckpt = copy.deepcopy(ckpt)
+
+    for key in keys:
+        if 'mlp.fc2' in key:
+            value = new_ckpt['state_dict'][key]
+            value = torch.cat([value, experts[key.replace('fc2.', f'experts.{target_expert}.')]], dim=0)
+            new_ckpt['state_dict'][key] = value
+
+    torch.save(new_ckpt, os.path.join(args.targetPath, 'coco.pth'))
+
+    names = ['aic', 'mpii', 'ap10k', 'apt36k','wholebody']
+    num_keypoints = [14, 16, 17, 17, 133]
+    weight_names = ['keypoint_head.deconv_layers.0.weight', 
+                    'keypoint_head.deconv_layers.1.weight', 
+                    'keypoint_head.deconv_layers.1.bias', 
+                    'keypoint_head.deconv_layers.1.running_mean', 
+                    'keypoint_head.deconv_layers.1.running_var', 
+                    'keypoint_head.deconv_layers.1.num_batches_tracked', 
+                    'keypoint_head.deconv_layers.3.weight', 
+                    'keypoint_head.deconv_layers.4.weight', 
+                    'keypoint_head.deconv_layers.4.bias', 
+                    'keypoint_head.deconv_layers.4.running_mean', 
+                    'keypoint_head.deconv_layers.4.running_var', 
+                    'keypoint_head.deconv_layers.4.num_batches_tracked', 
+                    'keypoint_head.final_layer.weight', 
+                    'keypoint_head.final_layer.bias']
+    
+    exist_range = True
+
+    for i in range(5):
+
+        new_ckpt = copy.deepcopy(ckpt)
+
+        target_expert = i + 1
+
+        for key in keys:
+            if 'mlp.fc2' in key:
+                expert_key = key.replace('fc2.', f'experts.{target_expert}.')
+                if expert_key in experts:
+                    value = new_ckpt['state_dict'][key]
+                    value = torch.cat([value, experts[expert_key]], dim=0)
+                else:
+                    exist_range = False
+
+                new_ckpt['state_dict'][key] = value
+
+        if not exist_range:
+            break
+
+        for tensor_name in weight_names:
+            new_ckpt['state_dict'][tensor_name] = new_ckpt['state_dict'][tensor_name.replace('keypoint_head', f'associate_keypoint_heads.{i}')]
+
+        for tensor_name in ['keypoint_head.final_layer.weight', 'keypoint_head.final_layer.bias']:
+            new_ckpt['state_dict'][tensor_name] = new_ckpt['state_dict'][tensor_name][:num_keypoints[i]]
+
+        torch.save(new_ckpt, os.path.join(args.target, f'{names[i]}.pth'))
+
+if __name__ == '__main__':
+    main()
\ No newline at end of file