Pedestrian Detection Algorithm Based on Improved YOLOv4

Pedestrian detection aims to automatically identify and locate pedestrian objects in images or videos, enabling applications such as intelligent surveillance, traffic safety, and crowd counting. To address the issue of low accuracy in pedestrian detection using the YOLOv4 object detection algorithm, an improved YOLOv4 algorithm is proposed to enhance pedestrian detection performance. By incorporating the DenseNet model and ECANet attention mechanism, experiments are conducted on the INRIA dataset. The improved YOLOv4 algorithm achieves an average precision (AP) of 93.97%, which is a 4.85% improvement compared to the original YOLOv4 algorithm.