Pedestrian Detection Algorithm Based on Improved YOLO v5

Aiming at the problems such as large scale change and poor detection accuracy, an improved YOLOv5 pedestrian detection algorithm based on drone view was proposed.Firstly, a multi-scale bidirectional feature network structure was proposed to enhance the multi-scale feature detection ability of YOLOv5 network, so as to adapt to pedestrian change detection at different scales.Secondly, a location attention detection layer is established to suppress background interference. Finally, the occlusion screening module is created to screen out the occluded pedestrians and improve the detection accuracy of the algorithm.In order to verify the performance of the algorithm, a comparison experiment is carried out with the existing pedestrian detection algorithm. The experimental results show that compared with the original YOLOv5 algorithm, the algorithm's accuracy is improved by 10.01%, and compared with the YOLOv5-tiny pedestrian detection algorithm, the algorithm's accuracy is improved by 5.74%, which can detect pedestrians well in the perspective of drones.