A Deep Learning Based Freestyle Swimming Stroke Posture Analysis and Feedback System

Freestyle swimming stroke is the basic yet the most technically challenging stroke in swimming. Proper technique is critical to minimize the risk of overuse injuries, mainly seen in the shoulders, lower back, and knees. This research paper presents a deep learning-based system that utilizes Long Short-Term memory (LSTM) networks to automate analysis of freestyle swimming stroke posture. The system evaluates swimmers' videos, identifying key joints and movements. By analyzing sequential movements, the LSTM model determines whether the swimmer's posture is correct or incorrect which lowers the risk of injury and improves the swimmer performance. The model was trained with a curated dataset of both correct and incorrect swimming poses, achieving an overall accuracy of 85%, demonstrating a strong performance in classifying stroke correctness. This study demonstrates a significant advancement in sports technology by integrating deep learning techniques for stroke posture analysis.