JAMCEN: A Gamified Fitness Application Using Real-Time Pose Estimation

Abstract

The growing demand for engaging, accessible fitness solutions has driven the integration of digital technologies with physical activity. However, many existing systems rely on complex machine learning models or specialized hardware, which may limit accessibility and interpretability. This study presents JAMCEN, a gamified fitness application that integrates real-time MediaPipe pose estimation and rule-based movement recognition to transform physical exercises into interactive gameplay. The study aimed to develop and evaluate a system that detects user movements and supports engagement through gamification without relying on computationally intensive models. The system was implemented as a mobile and web-based application integrating camera-based pose estimation, threshold-based movement interpretation, and game-driven interaction. We developed two exercise-based games, ZomJack and Splashy Fish, to represent full-body and upper-body movement activities. Evaluation included quantitative assessment of movement-detection accuracy and a user evaluation of 30 respondents focused on usability and engagement. The system achieved movement detection accuracies of 99.26% for jumping jacks and 99.31% for lateral arm raises. User evaluation yielded ratings of 5.0/5.0 for ease of use, 4.5/5.0 for responsiveness, and 4.6/5.0 each for engagement and clarity of feedback. Gameplay metrics also documented user interaction during the evaluated sessions. These findings suggest that combining real-time pose estimation, rule-based movement recognition, and gamification can support an interpretable and accessible interactive fitness experience. JAMCEN demonstrates the potential of user-centered, AI-assisted fitness applications to integrate physical exercise with digital gameplay.