DeciWatch: A Simple Baseline for 10× Efficient 2D and 3D Pose Estimation

This repo is the official implementation of "DeciWatch: A Simple Baseline for 10× Efficient 2D and 3D Pose Estimation". [Paper] [Project]

Update

• Clean version is released! It currently includes code, data, log and models for the following tasks:
• 2D human pose estimation
• 3D human pose estimation
• Body recovery via a SMPL model

TODO

• Provide different sample interval checkpoints/logs
• Add DeciWatch in MMHuman3D

Description

This paper proposes a simple baseline framework for video-based 2D/3D human pose estimation that can achieve 10 times efficiency improvement over existing works without any performance degradation, named DeciWatch. Unlike current solutions that estimate each frame in a video, DeciWatch introduces a simple yet effective sample-denoise-recover framework that only watches sparsely sampled frames, taking advantage of the continuity of human motions and the lightweight pose representation. Specifically, DeciWatch uniformly samples less than 10% video frames for detailed estimation, denoises the estimated 2D/3D poses with an efficient Transformer architecture, and then accurately recovers the rest of the frames using another Transformer-based network. Comprehensive experimental results on three video-based human pose estimation, body mesh recovery tasks and efficient labeling in videos with four datasets validate the efficiency and effectiveness of DeciWatch.

Getting Started

Environment Requirement

DeciWatch has been implemented and tested on Pytorch 1.10.1 with python >= 3.6. It supports both GPU and CPU inference.

Clone the repo:

git clone https://github.com/cure-lab/DeciWatch.git

We recommend you install the requirements using conda:

# conda
source scripts/install_conda.sh

Prepare Data

All the data used in our experiment can be downloaded here.

Google Drive

Baidu Netdisk

Valid data includes:

Please refer to doc/data.md for detailed data information and data preparing.

Training

Run the commands below to start training:

python train.py --cfg [config file] --dataset_name [dataset name] --estimator [backbone estimator you use] --body_representation [smpl/3D/2D] --sample_interval [sample interval N]

For example, you can train on 3D representation of 3DPW using backbone estimator SPIN with sample interval 10 by:

python train.py --cfg configs/config_pw3d_spin.yaml --dataset_name pw3d --estimator spin --body_representation 3D --sample_interval 10

Note that the training and testing datasets should be downloaded and prepared before training.

You may refer to doc/training.md for more training details.

Evaluation

Results on 2D Pose

Results on 3D Pose

Results on SMPL

Noted that although our main contribution is the efficiency improvement, using DeciWatch as post processing is also helpful for accuracy and smoothness improvement.

You may refer to doc/evaluate.md for evaluate details.

Quick Demo

Run the commands below to visualize demo:

python demo.py --cfg [config file] --dataset_name [dataset name] --estimator [backbone estimator you use] --body_representation [smpl/3D/2D] --sample_interval [sample interval N]

You are supposed to put corresponding images with the data structure:

|-- data
    |-- videos
        |-- pw3d 
            |-- downtown_enterShop_00
                |-- image_00000.jpg
                |-- ...
            |-- ...
        |-- jhmdb
            |-- catch
            |-- ...
        |-- aist
            |-- gWA_sFM_c01_d27_mWA2_ch21.mp4
            |-- ...
        |-- ...

For example, you can train on 3D representation of 3DPW using backbone estimator SPIN with sample interval 10 by:

python demo.py --cfg configs/config_pw3d_spin.yaml --dataset_name pw3d --estimator spin --body_representation 3D --sample_interval 10

Please refer to the dataset website for the raw images. You may change the config in lib/core/config.py for different visualization parameters.

You may refer to doc/visualize.md for visualization details.

Citing DeciWatch

If you find this repository useful for your work, please consider citing it as follows:

@article{zeng2022deciwatch,
  title={DeciWatch: A Simple Baseline for 10x Efficient 2D and 3D Pose Estimation},
  author={Zeng, Ailing and Ju, Xuan and Yang, Lei and Gao, Ruiyuan and Zhu, Xizhou and Dai, Bo and Xu, Qiang},
  journal={arXiv preprint arXiv:2203.08713},
  year={2022}
}

Please remember to cite all the datasets and backbone estimators if you use them in your experiments.

Acknowledgement

Many thanks to Xuan Ju for her great efforts to clean almost the original code!!!

License

This code is available for non-commercial scientific research purposes as defined in the LICENSE file. By downloading and using this code you agree to the terms in the LICENSE. Third-party datasets and software are subject to their respective licenses.