Development of a Real-Time Workout Validation and Repetition Counting Application Using YOLO-Based Human Pose Estimation

Maintaining physical fitness through regular workouts is essential for health; however, the fast-paced modern lifestyle often makes it difficult for individuals to work out consistently. Barriers such as limited time, the cost of gym memberships, and access to fitness facilities have led many people to choose home workouts as a more flexible alternative. Nevertheless, working out independently also presents challenges, such as difficulty in maintaining proper technique and accurately counting repetitions without a trainer's guidance. To address these issues, this study develops a machine learning-based workout application that integrates human pose estimation and classification models to validate movements and count exercise repetitions automatically and in real time. The system employs the YOLO algorithm for body detection and keypoint tracking to analyze movements during physical workouts such as squats and planks. Experimental results show that the application can recognize body poses with high accuracy and provide immediate feedback, thus improving the effectiveness of the workout and reducing the risk of injury. This application is expected to be a practical and efficient solution to support home workouts with guidance comparable to that of a personal trainer.