Khoi Tran Dinh Anh

Khoi Tran Dinh Anh

AI Learning

Tennis Game

Developed a comprehensive Tennis Game Analysis system that integrates YOLO for real-time player detection and TrackNet for precise tennis ball tracking, enabling detailed performance analysis and game statistics in live matches.

  • Player Detection with YOLO: Leveraged a fine-tuned YOLOv8 model to detect and identify players on the court, accurately capturing their positions, movements, and actions during the game. The model was trained to handle various camera angles, lighting conditions, and occlusions.

  • Tennis Ball Tracking with TrackNet: Implemented the TrackNet deep learning model to track the trajectory of the tennis ball in real-time. The system was optimized to follow the high-speed movement of the ball across the court and predict its location frame by frame, even during fast rallies and serves.

  • Game Performance Metrics: Combined player and ball tracking data to generate meaningful insights, such as player movement patterns, ball speed, shot accuracy, and court coverage. The system also provides a visual overlay of ball trajectories and heatmaps of player movements.

  • Real-time Application: Designed for real-time use in match analysis, training sessions, and broadcasting, providing instant feedback on player performance and ball tracking.

  • Technologies Used: Python, OpenCV, YOLOv5/YOLOv8, TrackNet, Pytorch, custom data visualization tools for real-time feedback.

This project demonstrates strong capabilities in computer vision, sports analytics, and deep learning applied to tracking and analyzing complex movements in a fast-paced sports environment.