README.md

Multi-view Human Action Recognition

Docker

Docker Compose for multi-view video streaming in ROS

cd docker
docker-compose up --build 

Unity

Unity project for a mixed reality environment

Scripts

Data preparation for mmpose and mmskeletion

1. split original video to action videolets

   python scripts/split_video.py

2. generates json for mmskeleton

   # this will generate model/florence_ccam.json
   python scripts/json_generation.py
   

Action Recognition Model

Deep learning models for action recognition

Test codes

Wookjin Choi [email protected]