Automated AI-Driven Highlights Generation for Football Matches

Abstract

This research work presents an AI-powered football highlight generation application that automates the process of detecting key events in a match using deep learning. Traditional methods of highlight creation are time-consuming and rely on human effort, often delaying access to important moments of the game. To address this limitation, the proposed system utilizes the YOLOv8 object detection algorithm to identify and track essential elements such as players, the ball, and goalposts. The proposed system automatically detects football match events and generates highlight reels through a web-based platform. Users upload a full match video, which is then processed to detect significant events, particularly goals, by analyzing the interaction between the ball and the goal line. A window of 125 frames before and 125 frames after the detected event is extracted to generate a 10-second clip for each event. These clips are stored and compiled into a highlight reel. These highlight reels can be viewed and downloaded through a user-friendly web interface. The research work demonstrates the potential of deep learning in sports highlights generation and analysis by providing a fast, accurate, and scalable solution for automatic video summarization.