# OpenPose: Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields

**Authors:** Zhe Cao, Gines Hidalgo, Tomas Simon, Shih-En Wei, Yaser Sheikh  
**arXiv:** [1812.08008](https://arxiv.org/abs/1812.08008)  
**Date:** 2018-12-18  
**Summary page:** https://www.summarizepaper.com/en/paper/1812.08008v1/

## Key points

- OpenPose is a system for real-time multi-person 2D pose detection.
- It includes body, foot, hand and facial keypoints.
- It has been developed by refining previous work on Part Affinity Fields (PAFs) and body part location estimation simultaneously across training stages.
- This results in a substantial increase in both runtime performance and accuracy.
- The authors have presented the first combined body and foot keypoint detector based on an internal annotated foot dataset that they have publicly released.
- The combined detector reduces inference time compared to running them sequentially while maintaining the accuracy of each component individually.
- OpenPose is the first open-source real-time system for multi-person 2D pose detection regardless of the number of people in the image.
- The release of OpenPose marks a significant milestone in enabling machines to have an understanding of people in images and videos with high accuracy and real-time performance.

## Summary

OpenPose is a groundbreaking system for real-time multi-person 2D pose detection that includes body, foot, hand and facial keypoints. It has been developed by refining previous work on Part Affinity Fields (PAFs) and body part location estimation simultaneously across training stages. This results in a substantial increase in both runtime performance and accuracy. The authors also present the first combined body and foot keypoint detector based on an internal annotated foot dataset that they have publicly released. They show that the combined detector not only reduces inference time compared to running them sequentially but also maintains the accuracy of each component individually. OpenPose is the first open-source real-time system for multi-person 2D pose detection regardless of the number of people in the image. The release of OpenPose marks a significant milestone in enabling machines to have an understanding of people in images and videos with high accuracy and real-time performance.