FitVerse: AI-Driven Real-Time Pose Correction and Fitness Recommendation System

Abstract

Physical fitness and posture play an essential role in maintaining human health and preventing injuries. However, the rapid growth of digital lifestyles has reduced physical activity and increased posture-related disorders. Modern fitness applications offer tracking capabilities but lack real-time feedback, personalized adaptation, and integration of artificial intelligence for individualized guidance. This paper introduces FitVerse, an AI-driven real-time pose correction and fitness recommendation system designed to provide browser-based exercise feedback, progress tracking, and diet management. FitVerse integrates MediaPipe and MoveNet for real-time pose estimation, combined with Generative AI and Random Forest-based machine learning models to deliver personalized workout and nutritional recommendations. Additionally, the system includes a food calorie tracker and a user progress dashboard, improving motivation, goal achievement, and injury prevention. Experimental evaluation demonstrates FitVerse’s capability to perform accurate pose detection, generate context-aware exercise advice, and maintain low inference latency even in browser-based environments. FitVerse offers a scalable, lightweight, and user-centric AI fitness platform that combines pose correction, recommendation systems, and health analytics in a unified architecture.