hand clapping recognition

develop an imitation detection system by implementing a human action recognition (HAR) algorithm using deep learning. the project is to use robot-based interaction to help ASD children enhancing their imitation skills. each child will join multiple therapy sessions with NAO robot include imitation scenario (hand clapping). All sessions are recorded via robot’s camera and used in the imitation detection system. Video preprocessing, feature extraction, and classification will be applied to the recordings to determine whether a child imitates correctly or not.

So, I want to build a deep learning model (human action recognition) using Python, that process videos and detect hand clapping. There is open dataset called UCF50.