Abstract: In the past years, thousands of fitness enthusiasts are seeking solutions to having an effective and personalized workout. Presently, most of the users experience a bad posture during some exercises that results in pain or reduced outcome. To overcome this challenge, the "AI-Based Workout Guide" is a technology-driven solution which uses computer vision and AI to give real-time posture correction along with rep counting.
The project uses AI and machine learning algorithms to assess the motion of users' bodies while executing a workout. Captured through a camera or smartphone, the system will compare this posture against predefined models for the best techniques to do the exercises. If there is an incorrect posture, it will always give instant feedback as to what adjustments are needed. Another system feature is that the number of repetitions is automatically counted, so there's no manual counting involved and even more concentration on the proper form by the user.
The model has been trained on the dataset of various exercise postures, including squats, push-ups, and lunges. Computer vision libraries OpenPose or Mediapipe are used to detect key landmarks, such as joint angles and alignment. In real time, the system evaluates these landmarks in order to provide an accurate posture assessment and count of the rep. In the end, this would help the users improve their workout efficiency, reduce the chances of injury, and achieve fitness goals in a better manner.
It's an accessible, scalable AI-based workout guide, from which follows that it can easily be adapted into any mobile or web application. Its applicability extends to novices and more experienced fitness enthusiasts. This project shows how technology may transform personal training in fitness: a combination of advanced AI techniques with a practical solution for the application of fitness.
Keywords: AI, Computer Vision, Workout Guide, Pose Estimation, MediaPipe, OpenCV, Exercise Form Correction.
|
DOI:
10.17148/IJARCCE.2024.131011
