Real-Time Human Tracking and Activity Recognition using GPS, IPS, RFID, and Computer Vision

Tracking and understanding human behaviour in real time is crucial across surveillance, safety, and smart environment applications. This study aims to create an integrated framework that seamlessly localises, identifies, and monitors individuals indoors and outdoors. The methodology fuses Global Positioning System (GPS), Indoor Positioning System (IPS), Radio Frequency Identification (RFID), and video surveillance to accurately localise, identify, and monitor individuals both indoors and outdoors. GPS offers reliable outdoor localisation, while IPS and RFID collectively ensure seamless indoor tracking. Person detection is performed using the YOLOv8 model, identity recognition is achieved through FaceNet with a trained Support Vector Machine (SVM), and activity recognition is handled by a CNN-LSTM architecture. The system processes live video or uploaded footage, presenting real-time information such as name, location coordinates, and detected activity through a user-friendly interface. The demonstrated fusion of multi-sensor localisation and vision analytics yields a low-cost, robust, and scalable platform for real-time human tracking and behaviour analysis, outperforming single-sensor solutions and enabling intelligent monitoring across diverse real-world settings.