Human Recognition

1. Introduction

First, you need to train an YOLOv8-pose model or leverage pretrained weights. Then export the weights file to onnx format following the instruction here. I recommend using the pretrained model YOLOv8l-pose because it is accurate enough although a bit slow.
Second, it's also necessary to prepare a video for infering.
After that, you can run the following command and watch the result:

python demo.py -iv demo.mp4

Name		Name	Last commit message	Last commit date
Latest commit History 19 Commits
JHMDB		JHMDB
assets		assets
weights		weights
yolov8-pose		yolov8-pose
README.md		README.md
demo.py		demo.py
requirements.txt		requirements.txt
utils_action.py		utils_action.py
utils_pose.py		utils_pose.py