This repository contains the code and models for an action recognition system that classifies actions into three categories: Boxing, Kick, and Nonfighting. The system is part of a pipeline to integrate action recognition into Unity using Unity Sentis.
Git for Unity project related to the pipeline: https://github.com/ArianMoeini/UnityDone
- DifferentSavedData/: Directory for various saved datasets.
- Logs/: Directory for storing log files.
- MP_Data/: Directory for storing MediaPipe data.
- ActionDetection.ipynb: Jupyter notebook containing the main code for action detection.
- README.md: This file.
- extraData.zip: Additional data for the project.
- mainModel.h5: The main model file in H5 format.
- model.onnx: The model file in ONNX format.
- model2.h5: Another model file in H5 format.