PHYSIOTHERAPY POSE DETECTION MODEL

Human pose recognition has become an important focus in computer vision, particularly for applications in physiotherapy and self-assessment. In this work, we present an approach for accurate physiotherapy pose assessment using deep learning algorithms. The proposed system utilizes pose detection to facilitate the self-guided learning of physiotherapy exercises. Specifically, our approach employs multi-part pose detection through a standard PC camera to capture and analyze physiotherapy poses in real-time. We introduce an enhanced scoring algorithm capable of assessing various poses, which ensures adaptability across different pose types and environments. Additionally, a hybrid machine learning model is implemented using Linear Regression to extract features from key points identified in each frame, leveraging the Open Pose framework. The robustness of this system is evaluated across multiple physiotherapy poses in diverse settings, demonstrating its potential for effective, real-time physiotherapy assessment