tyicestream / lightweight-human-pose-estimation.opencv

OpenCV inference implementation of Paper "Real-time 2D Multi-Person Pose Estimation on CPU: Lightweight OpenPose "(https://arxiv.org/abs/1811.12004)

lightweight-human-pose-estimation.OpenCV

This repository contains the inference code for the paper Real-time 2D Multi-Person Pose Estimation on CPU: Lightweight OpenPose.

We use the OpenCV library for the model inference, not including other library.

We use the model from lightweight-human-pose-estimation.pytorch and rewrite the post processing code in c++ with OpenCV.

Requirements

• Ubuntu 20.04
• OpenCV (compile with CUDA)

Demo

We provide demo just for the quick results preview. We only use one picture for show the result. If you want, you can use the algorithm for videos or webcam.

• mkdir build && cd build
• cmake .. && make && ./poseEstimation

Citation:

If this helps your research, please cite the paper:

@inproceedings{osokin2018lightweight_openpose,
    author={Osokin, Daniil},
    title={Real-time 2D Multi-Person Pose Estimation on CPU: Lightweight OpenPose},
    booktitle = {arXiv preprint arXiv:1811.12004},
    year = {2018}
}