Description
Motion analysis, gesture interaction and animation all start with finding a person's joints in the frame. OpenPose from Carnegie Mellon is the first real-time system to jointly detect body, hand, face and foot keypoints for multiple people, up to 135 keypoints per image.
Written in C++ for Linux, macOS and Windows, it handles images, video and webcams and is a widely cited foundation in pose estimation.
Whole-body keypoints:135 points across body, hands, face and feet.
Multi-person real time:Many people per frame.
Inputs:Images, video and webcams.
Cross-platform:Builds on Linux, macOS and Windows.
Written in C++ for Linux, macOS and Windows, it handles images, video and webcams and is a widely cited foundation in pose estimation.
Features
Whole-body keypoints:135 points across body, hands, face and feet.
Multi-person real time:Many people per frame.
Inputs:Images, video and webcams.
Cross-platform:Builds on Linux, macOS and Windows.
Tags:computer-vision