Human Pose Estimation Using Deep Neural Network

In this research, we present an innovative system for human pose estimation designed to recognize and provide feedback on exercises such as push-ups, squats, and planks. Leveraging Mediapipe and Python, our model captures real-time or uploaded videos to detect postures, measure body angles, and count repetitions. The system evaluates exercise correctness, ensuring users maintain proper form through detailed visual feedback. By identifying key body points and calculating angles, our model distinguishes between correct and incorrect postures, providing real-time corrections. The implementation involves stages like data acquisition, preprocessing, pose estimation, and feedback generation. Mediapipe’s pose detection algorithm tracks key body landmarks, while Python scripts analyze these to determine angles and movements. This data counts repetitions and evaluates form, offering immediate visual feedback.Our solution demonstrates high accuracy in recognizing and evaluating exercises, proving valuable for fitness enthusiasts and professionals. Extensive testing shows the system accurately counts repetitions and identifies improper form, enhancing workout routines. This research highlights the potential of pose estimation technologies to revolutionize the fitness industry, ensuring proper form, preventing injuries, and maximizing exercise effectiveness.