Frequent physical activities may significantly improve someone’s health and well-being, but if not done properly, they can also be useless and even dangerous. A frequent error committed by users might be inappropriate exercise posture, which results from the wrong form or posture. This study introduces Exercise Trainer, a system that analyses a user’s posture during exercise and offers tailored advice on how to improve form. The system collects motion data from users using an optical camera, assesses their posture, contrasts it with ideal postures, and then offers suggestions for improving their posture. The suggested model makes use of the TensorFlow MoveNet model, which forecasts 17 key points or coordinates of different body parts using the JSON file produced by TensorFlow.js. TensorFlow.js generates a JSON file that interfaces with the web application, enabling the system to provide the user with real- time feedback on the correctness of their workout posture. This feature makes sure that the user gets instant feedback so they may correct their form as needed while exercising.
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.