This project addresses the challenge of gym overcrowding while mitigating the privacy concerns associated with biometric scanning.
By leveraging a RealSense depth camera, the system periodically captures spatial data of the gym environment, converting it into structured CSV datasets. These datasets are then processed by a machine learning pipeline, which dynamically analyzes machine occupancy trends to generate optimized workout schedules.
Users receive personalized low-wait-time workout plans, guiding them on the best machines to use and the ideal time to visit the gym for minimal congestion. The system employs computer vision-based object recognition to identify gym machines and individuals within the environment without requiring intrusive biometric data collection.
Instead of tracking individuals directly, it focuses on spatiotemporal occupancy patterns, proving that efficient gym management can be achieved without compromising personal privacy.
Future iterations will incorporate user-selected timeframes, but the current implementation is optimized for a rolling two-hour dataset. This project is an advanced take on privacy-conscious real-time gym space optimization using depth sensing and AI-driven predictive modeling.