FitGym: An Online Smart GYM Platform

This project presents a Smart Online Fitness Platform, an intelligent web-based solution designed to transform traditional fitness training by integrating artificial intelligence, computer vision, and personalized health management into a single unified platform. The system aims to provide users with a comprehensive digital fitness experience by combining personalized workout planning, nutrition guidance, real-time exercise monitoring, and interactive virtual coaching. Through a user-friendly interface, individuals can access customized fitness programs based on their personal goals, fitness levels, and progress, enabling them to maintain a structured and effective workout routine from any location. The platform incorporates a secure authentication system that allows users, trainers, and administrators to access role-specific functionalities. Certified trainers can create and assign personalized workout schedules, monitor user performance, update exercise plans, and provide professional guidance remotely. A dynamic workout calendar and task management module help users organize their daily exercise routines, set fitness goals, receive reminders, and track completed activities, promoting consistency and long-term engagement. To improve user motivation and retention, the platform integrates a comprehensive gamification framework consisting of achievement badges, experience points, reward levels, progress milestones, and interactive leader boards. Users earn rewards by completing workout sessions, maintaining exercise streaks, and achieving fitness objectives, creating an engaging and competitive environment that encourages continuous participation. In addition, detailed progress analytics enable users to monitor improvements in workout completion, calorie expenditure, body measurements, and overall fitness performance through intuitive graphical dashboards. The platform also provides personalized dietary recommendations by generating nutrition plans that complement individual fitness goals, whether focused on weight loss, muscle gain, endurance improvement, or general wellness. These recommendations are continuously updated according to user progress and workout intensity, ensuring balanced nutritional support throughout the fitness journey. A conversational AI chatbot is integrated into the platform to provide instant assistance and guidance. The chatbot answers fitness-related questions, explains exercise techniques, offers workout recommendations, assists with navigation throughout the application, and provides immediate support whenever users require help. This intelligent assistant enhances accessibility and improves the overall user experience by delivering real-time responses without requiring human intervention. One of the major innovations of the platform is the integration of an AI-powered Transformer-based Yoga Pose Detection and Correction System. The proposed Residual Biomechanical Transformer Network (RBT-Net) combines YOLOv8m-Pose for accurate human pose estimation with Transformer-based temporal reasoning and a ResidualMLP backbone to analyze body movements across sequential frames. Instead of relying solely on individual images, the Transformer encoder captures temporal dependencies between consecutive poses, enabling the system to understand movement progression, pose stability, and posture transitions more effectively. The extracted biomechanical features, including normalized skeletal coordinates, joint angles, limb distances, and confidence scores, are processed through the hybrid architecture to accurately classify yoga poses while identifying posture deviations. Beyond pose recognition, the intelligent correction framework estimates posture severity, detects joint-level errors, and generates personalized corrective feedback based on the user’s current posture and skill level. This allows the platform to function as a virtual yoga instructor capable of providing real-time guidance that helps users improve posture accuracy, reduce injury risks, and perform yoga exercises more safely and effectively. The integration of Transformer-based temporal modeling significantly enhances robustness against body movement variations and improves recognition performance during dynamic yoga sessions compared to conventional frame-based approaches. Overall, the Smart Online Fitness Platform successfully combines personalized fitness management, intelligent workout planning, nutrition guidance, gamification, conversational AI, and advanced Transformer-based yoga posture analysis into a unified digital ecosystem. By integrating modern artificial intelligence techniques with interactive web technologies, the platform delivers a comprehensive, accessible, and engaging fitness solution that bridges the gap between conventional gym environments and next-generation intelligent virtual fitness systems.