HEViTPose: towards high-accuracy and efficient 2D human pose estimation with cascaded group spatial reduction attentio

Code coming soon...

Paper

[HEViTPose: towards high-accuracy and efficient 2D human pose estimation with cascaded group spatial reduction attention]

Chengpeng Wu

The network architecture of HEVITPose

Main Results

With the code contained in this repo, you should be able to reproduce the following results.

Results on MPII val and test set

Method	Test set	Input size	Params	GFLOPs	Hea	Sho	Elb	Wri	Hip	Kne	Ank	Total
HEViTPose-T	MPII val	256×256	3.21M	1.75G	95.9	94.9	87.4	81.6	87.4	81.6	77.2	87.2
HEViTPose-S	MPII val	256×256	5.88M	3.64G	96.3	95.2	88.7	83.3	88.5	83.9	79.5	88.5
HEViTPose-B	MPII val	256×256	10.63M	5.58G	96.5	95.6	89.5	84.5	89.1	85.7	81.1	89.4
HEViTPose-T	MPII test-dev	256×256	3.21M	1.75G	97.6	95.1	89.0	83.6	89.1	83.9	79.1	88.7
HEViTPose-S	MPII test-dev	256×256	5.88M	3.64G	97.8	95.9	90.5	86.0	89.7	86.0	81.7	90.1
HEViTPose-B	MPII test-dev	256×256	10.63M	5.58G	98.0	96.1	91.3	86.5	90.2	86.6	83.0	90.7

Results on COCO val2017 and test-dev2017 set

Method	Test set	Input size	AP	AP .5	AP .75	AP (M)	AP (L)	AR
HEViTPose-B	COCO val	256×256	75.4	93.6	83.5	72.4	79.6	78.2
HEViTPose-B	COCO test-dev	256×256	72.6	92.0	80.9	69.2	78.2	78.0

Visualization

Some examples of the prediction results of the HEViTPose network model for human posture include occlusion, multiple people, viewpoint and appearance change on the MPII (top) and COCO (bottom) data sets.

Installation

1. Clone code

    git clone https://github.com/T1sweet/HEViTPose
    cd ./HEViTPose

2. Create a conda environment for this repo

    conda create -n HEViTPose python=3.9
    conda activate HEViTPose

3. Install PyTorch >= 1.6.0 following official instruction, e.g.

Our model is trained in a GPU platforms and relies on the following versions: torch==1.10.1+cu113, torchvision==0.11.2+cu113

    conda install pytorch torchvision cudatoolkit=11.3 -c pytorch

4. Install other dependency python packages

Our code is based on the MMPose 0.29.0 code database, and dependencies can be installed through the methods provided by MMPose. Install MMCV using MIM.

    conda install pytorch torchvision cudatoolkit=11.3 -c pytorch
    pip install -U openmim
    mim install mmcv-full==1.4.5

Install other dependency.

    pip install -r requirements.txt

5. Prepare dataset

Download MPII and COCO from website and put the zip file under the directory following below structure, (xxx.json) denotes their original name.

./data
|── coco
│   └── annotations
|   |   └──coco_train.json(person_keypoints_train2017.json)
|   |   └──coco_val.json(person_keypoints_val2017.json)
|   |   └──coco_test.json(image_info_test-dev2017.json)
|   └── images
|   |   └──train2017
|   |   |   └──000000000009.jpg
|   |   └──val2017
|   |   |   └──000000000139.jpg
|   |   └──test2017
|   |   |   └──000000000001.jpg
├── mpii
│   └── annotations
|   |   └──mpii_train.json(refer to DEKR, link:https://github.com/HRNet/DEKR)
|   |   └──mpii_val.json
|   |   └──mpii_test.json
|   |   └──mpii_gt_val.mat
|   └── images
|   |   └──100000.jpg

Usage

1. Download trained model

• MPII
• COCO

2. Evaluate Model

Change the checkpoint path by modifying pretrained in HEViTPose-B_mpii_256x256.py, and run following commands: python tools/test.py config checkpoint config option means the configuration file, which must be set. checkpoint option means the training weight file and must be set.

# evaluate HEViTPose-B on mpii val set
python tools/test.py ../configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/HEViTPose-B_mpii_256x256.py /work_dir/HEViTPose/HEViTPose-B.pth

# evaluate HEViTPose-S on mpii val set
python tools/test.py ../configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/HEViTPose-S_mpii_256x256.py /work_dir/HEViTPose/HEViTPose-S.pth

# evaluate HEViTPose-T on mpii val set
python tools/test.py ../configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/HEViTPose-T_mpii_256x256.py /work_dir/HEViTPose/HEViTPose-T.pth

# evaluate HEViTPose-B on coco val set
python tools/test.py ../configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/HEViTPose-B_coco_256x256.py /work_dir/HEViTPose/HEViTPose-B_coco.pth

3. Train Model

Change the checkpoint path by modifying pretrained in HEViTPose-B_mpii_256x256.py, and run following commands:

# evaluate HEViTPose-B on mpii val set
python tools/train.py ../configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/HEViTPose-B_mpii_256x256.py

# evaluate HEViTPose-B on coco val2017 set
python tools/train.py ../configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/HEViTPose-B_coco_256x256.py

Contact me

If you have any questions about this code or paper, feel free to contact me at [email protected].

Citations

If you find this code useful for your research, please cite our paper:

@misc{wu2024hevitpose,
    title     = {HEViTPose: High-Efficiency Vision Transformer for Human Pose Estimation},
    author    = {Chengpeng Wu, Guangxing Tan*, Chunyu Li},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    year      = {2023},
    eprint={2311.13615 },
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}

Acknowledgement

This algorithm is based on code database MMPose, and its main ideas are inspired by EfficientViT, PVTv2, Swin and other papers.

@misc{mmpose2020,
    title={OpenMMLab Pose Estimation Toolbox and Benchmark},
    author={MMPose Contributors},
    howpublished = {\url{https://github.com/open-mmlab/mmpose}},
    year={2020}
}