Abstract

This research presents an innovative approach to pedestrian detection in video surveillance, leveraging the power of YOLOv5 (You Only Look Once version 5) combined with light perception fusion-based feature extraction. The proposed methodology aims to enhance the accuracy and efficiency of pedestrian detection systems in varying lighting conditions. YOLOv5, known for its real-time object detection capabilities, is integrated with a novel feature extraction technique that fuses information from multiple light perception sensors. This fusion strategy allows the model to adapt and perform robustly in diverse lighting scenarios. The experimental results demonstrate the superiority of the proposed method, achieving a remarkable performance. The fusion of YOLOv5 with light perception-based feature extraction showcases promising advancements in pedestrian detection, addressing challenges posed by dynamic lighting conditions in real-world surveillance environments.