Smart Gym Assistant: Posture Monitoring and Rep Counting

• Published in: International Conference on Signal Processing and Communication (Online) pp. 568 - 574
• Main Authors: Jayasankar, Arya, Uday, Krishnendu M, Shokeen, Ishaan, Lalitha, S.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 20.02.2025
• Subjects: Convolution Neural Network; deep learning models; Injuries; Machine learning algorithms; MediaPipe; Monitoring; Privacy; real-time monitoring; Real-time systems; Schedules; Testing; Training; Tutorials; Videos
• ISSN: 2643-444X
• DOI: 10.1109/ICSC64553.2025.10968414

Many workout applications offer their users a set of videos to follow, where users have no opportunity to have their form corrected in real time. This approach can lead to unnatural alignment, increase the risk of injury and diminish the usefulness of the exercises. The workout pose corrector with an attached integrated repetition counter is the focus of this paper for improving performance and reducing the possibility of injury while exercising. The system employs state-of-the-art computer vision technology and machine learning algorithms to record video and analyze real-time body position during exercises. It provides error feedback information to the user from the evaluation of skeletal points against that of the ideal pose of a specific exercise. It also contains a counter mechanism which is activated only when the exercise is performed correctly. The information from the experiments revealed that the pose detection system was trained and tested with a precision of 98. 3% during training and 96. 4% during testing, while guaranteeing that the users perform the necessary exercises in the right form.