Posture identification for Tennis video

Deployed a deep learning pipeline combining YOLO and LSTM to analyze tennis shots and generate advanced stats from video footage. YOLO detected players, rackets, and the ball, while an LSTM classified shot types based on movement sequences. Object tracking with court projection enabled precise positioning analysis. Heatmaps visualized player movement and shot distribution. The pipeline ran on a Google Cloud VM, delivering real-time insights via an interactive Streamlit interface for in-depth match analysis