Person Detection using Proposed YOLOv8 algorithm

Surveillance systems are vital for detecting and preventing crime, especially in urban environments where real-time monitoring is crucial. This research proposes an advanced surveillance system that combines proposed YOLOv8 object detection with DeepSORT tracking for tracking a person in crowded place, enhanced with a DenseNet backbone, achieves 97.1% detection accuracy and an mAP of 0.971 at an IoU of 0.5, outperforming YOLOv3 and YOLOv5. The model makes use of Spatial Pyramid Pooling (SPP) and Feature Pyramid Networks (FPN) for multi-scale detection and an anchor-free approach for improved accuracy. Integrated with DeepSORT, the system tracks individuals reliably across frames, even in dynamic environments. Experimental results demonstrate that this approach significantly improves detection and tracking efficiency, making it ideal for automated surveillance and crime prevention in urban areas.