Football Analysis

• Tech Stack: Python, YOLO, Ultralytics, OpenCV, Clustering, Pandas, NumPy
• Github URL: Project Link

Developed an AI-based football analytics system to detect and track players, referees, and the ball in match videos using the YOLO object detection model.

Trained the model for improved accuracy and used K-Means clustering for pixel-based segmentation to classify players into teams based on t-shirt colors, enabling calculation of team ball possession percentages.

Implemented Optical Flow to estimate camera movement and Perspective Transformation to map scene depth, allowing accurate player movement measurement in meters and computation of individual speed and distance covered.