Back to projects
Nov 20, 2023
1 min read

Video-Based Injury Classification and Prevention in Football using Machine Learning

Developed a machine learning system to predict football player injuries from video analysis using advanced pose estimation and classification algorithms.

Successfully employed MediaPipe Pose and machine learning algorithms to predict football player injuries based on video analysis. Extracted relevant features from pose landmarks, including joint angles, limb lengths, and velocity vectors, to train machine learning models such as K-Means, K-NN, SVM, and Random Forest for injury prediction. Evaluated classifier performance using metrics like accuracy, precision, recall, and F1 score, and fine-tuned models to enhance predictive accuracy.

The project aims to improve injury prevention and player safety by providing early warnings based on analysis of player movements and poses.