/RSN

"Learning Delicate Local Representations for Multi-Person Pose Estimation" (ECCV 2020 Spotlight) & (COCO 2019 Human Keypoint Detection Challenge Winner) & (COCO 2019 Best Paper Award)

Primary LanguagePythonMIT LicenseMIT

PWC

PWC PWC

PWC PWC

Learning Delicate Local Representations for Multi-Person Pose Estimation (ECCV 2020 Spotlight)

winner bpa arXiv zhihu

*This is a pytorch realization of Residual Steps Network which won 2019 COCO Keypoint Challenge and ranks 1st place on both COCO test-dev and test-challenge datasets as shown in COCO leaderboard.

News

  • 2020.09 : Our RSN has been integrated into the great MMPose framework. Thanks to their effort. Welcome to use their codebase with their pre-trained model zoo. ⭐
  • 2020.07 : Our paper has been accepted as Spotlight by ECCV 2020 🚀
  • 2019.09 : Our work won the First place and Best Paper Award in COCO 2019 Keypoint Challenge 🏆

Abstract: In this paper, we propose a novel method called Residual Steps Network (RSN). RSN aggregates features with the same spatialsize (Intra-level features) efficiently to obtain delicate local representations, which retain rich low-level spatial information and result in precise keypoint localization. In addition, we propose an efficient attention mechanism - Pose Refine Machine (PRM) to further refine the keypoint locations. Our approach won the 1st place of COCO Keypoint Challenge 2019 and achieves state-of-the-art results on both COCO and MPII benchmarks, without using extra training data and pretrained model. Our single model achieves 78.6 on COCO test-dev, 93.0 on MPII test dataset. Ensembled models achieve 79.2 on COCO test-dev, 77.1 on COCO test-challenge dataset.

Pipieline of Residual Steps Network

Overview of RSN.

Architecture of Pose Refine Machine

Overview of RSN.

Some prediction resullts of our method on COCO and MPII valid datasets

Prediction Results of COCO-valid.

Prediction Results of MPII-valid.

Results(Original Version)

Results on COCO val dataset

Model Input Size GFLOPs AP AP50 AP75 APM APL AR
Res-18 256x192 2.3 70.7 89.5 77.5 66.8 75.9 75.8
RSN-18 256x192 2.5 73.6 90.5 80.9 67.8 79.1 78.8
RSN-50 256x192 6.4 74.7 91.4 81.5 71.0 80.2 80.0
RSN-101 256x192 11.5 75.8 92.4 83.0 72.1 81.2 81.1
2×RSN-50 256x192 13.9 77.2 92.3 84.0 73.8 82.5 82.2
3×RSN-50 256x192 20.7 78.2 92.3 85.1 74.7 83.7 83.1
4×RSN-50 256x192 29.3 79.0 92.5 85.7 75.2 84.5 83.7
4×RSN-50 384x288 65.9 79.6 92.5 85.8 75.5 85.2 84.2

Results on COCO test-dev dataset

Model Input Size GFLOPs AP AP50 AP75 APM APL AR
RSN-18 256x192 2.5 71.6 92.6 80.3 68.8 75.8 77.7
RSN-50 256x192 6.4 72.5 93.0 81.3 69.9 76.5 78.8
2×RSN-50 256x192 13.9 75.5 93.6 84.0 73.0 79.6 81.3
4×RSN-50 256x192 29.3 78.0 94.2 86.5 75.3 82.2 83.4
4×RSN-50 384x288 65.9 78.6 94.3 86.6 75.5 83.3 83.8
4×RSN-50+ - - 79.2 94.4 87.1 76.1 83.8 84.1

Results on COCO test-challenge dataset

Model Input Size GFLOPs AP AP50 AP75 APM APL AR
4×RSN-50+ - - 77.1 93.3 83.6 72.2 83.6 82.6

Results on MPII dataset

Model Split Input Size Head Shoulder Elbow Wrist Hip Knee Ankle Mean
4×RSN-50 val 256x256 96.7 96.7 92.3 88.2 90.3 89.0 85.3 91.6
4×RSN-50 test 256x256 98.5 97.3 93.9 89.9 92.0 90.6 86.8 93.0

Results(Pytorch Version)

Results on COCO val dataset

Model Input Size GFLOPs AP AP50 AP75 APM APL AR
Res-18 256x192 2.3 65.2 87.3 71.5 61.2 72.2 71.3
RSN-18 256x192 2.5 70.4 88.8 77.7 67.2 76.7 76.5

Note

  • + means using ensemble models.
  • All models are trained on 8 V100 GPUs
  • We done all the experiments using our orginal DL-Platform, all results in our paper are reported on this DL-Platform. There are some differences between it and Pytorch.

Repo Structure

This repo is organized as following:

$RSN_HOME
|-- cvpack
|
|-- dataset
|   |-- COCO
|   |   |-- det_json
|   |   |-- gt_json
|   |   |-- images
|   |       |-- train2014
|   |       |-- val2014
|   |
|   |-- MPII
|       |-- det_json
|       |-- gt_json
|       |-- images
|   
|-- lib
|   |-- models
|   |-- utils
|
|-- exps
|   |-- exp1
|   |-- exp2
|   |-- ...
|
|-- model_logs
|
|-- README.md
|-- requirements.txt

Quick Start

Installation

  1. Install Pytorch referring to Pytorch website.

  2. Clone this repo, and config RSN_HOME in /etc/profile or ~/.bashrc, e.g.

export RSN_HOME='/path/of/your/cloned/repo'
export PYTHONPATH=$PYTHONPATH:$RSN_HOME
  1. Install requirements:
pip3 install -r requirements.txt
  1. Install COCOAPI referring to cocoapi website, or:
git clone https://github.com/cocodataset/cocoapi.git $RSN_HOME/lib/COCOAPI
cd $RSN_HOME/lib/COCOAPI/PythonAPI
make install

Dataset

COCO

  1. Download images from COCO website, and put train2014/val2014 splits into $RSN_HOME/dataset/COCO/images/ respectively.

  2. Download ground truth from Google Drive or Baidu Drive (code: fc51), and put it into $RSN_HOME/dataset/COCO/gt_json/.

  3. Download detection result from Google Drive or Baidu Drive (code: fc51), and put it into $RSN_HOME/dataset/COCO/det_json/.

MPII

  1. Download images from MPII website, and put images into $RSN_HOME/dataset/MPII/images/.

  2. Download ground truth from Google Drive or Baidu Drive (code: fc51), and put it into $RSN_HOME/dataset/MPII/gt_json/.

  3. Download detection result from Google Drive or Baidu Drive (code: fc51), and put it into $RSN_HOME/dataset/MPII/det_json/.

Log

Create a directory to save logs and models:

mkdir $RSN_HOME/model_logs

Train

Go to specified experiment repository, e.g.

cd $RSN_HOME/exps/RSN50.coco

and run:

python config.py -log
python -m torch.distributed.launch --nproc_per_node=gpu_num train.py

the gpu_num is the number of gpus.

Test

python -m torch.distributed.launch --nproc_per_node=gpu_num test.py -i iter_num

the gpu_num is the number of gpus, and iter_num is the iteration number you want to test.

Citation

Please considering citing our projects in your publications if they help your research.

@inproceedings{cai2020learning,
  title={Learning Delicate Local Representations for Multi-Person Pose Estimation},
  author={Yuanhao Cai and Zhicheng Wang and Zhengxiong Luo and Binyi Yin and Angang Du and Haoqian Wang and Xinyu Zhou and Erjin Zhou and Xiangyu Zhang and Jian Sun},
  booktitle={ECCV},
  year={2020}
}

@inproceedings{cai2019res,
  title={Res-steps-net for multi-person pose estimation},
  author={Cai, Yuanhao and Wang, Zhicheng and Yin, Binyi and Yin, Ruihao and Du, Angang and Luo, Zhengxiong and Li, Zeming and Zhou, Xinyu and Yu, Gang and Zhou, Erjin and others},
  booktitle={Joint COCO and Mapillary Workshop at ICCV},
  year={2019}
}