• Install requirements.txt
• Install OpenCV
• Install tkinter
• copy config_template.py to config.py and edit as required (for reference look at configs/base_config)

This is a PyTorch implementation of Realtime_Multi-Person_Pose_Estimation (origin code is here https://github.com/ZheC/Realtime_Multi-Person_Pose_Estimation and https://github.com/CMU-Perceptual-Computing-Lab/openpose) The post processing is a little different from OpenPose causing a little different results as well as a few performance problems.

Converted caffe weights: https://cogsys.reutlingen-university.de/pub/files/op_converted.pth