GBS-YOLOv5: Improved Algorithm for UAV Intelligent Transportation Based on YOLOv5

Abstract

As the road traffic situation becomes complex, the task of traffic management takes on an increasingly heavy load. The air-to-ground traffic administration network of drones has become an important tool to promote the high quality of traffic police work in many places. Drones can be used instead of a large number of human beings to perform daily tasks, as: traffic offense detection, daily crowd detection, etc. UAVs are less accurate at detecting drones because they operate from the air and take pictures of small targets. To address the problem of low accuracy of UAVs in detecting small targets, we proposed the GBS-YOLOv5 model. It was an improvement on the original YOLOv5 model. The default YOLOv5 model suffered from severe loss of small target information and under-utilization of shallow feature information as the depth of the feature extraction network deepened. We deepened the feature extraction network by introducing an efficient spatiotemporal interaction (ESI) module. We added the spatial pyramid convolution (SPC) module as a detection head for tiny-sized targets. A shallow bottleneck (SB) was introduced to better preserve the detailed information of small targets in the shallow features. The introduction of recursive gated convolution in the feature fusion section enabled better interaction of higher-order spatial semantic information. The GBS-YOLOv5 algorithm conducted experiments showing that the value of mAP@0.5 was 35.3% and the mAP@0.5:0.95 was 20.0%. Compared to the default YOLOv5 algorithm was boosted by 4.0% and 3.5% respectively.

A Method for Detecting Lightweight Optical Remote Sensing Images Using Improved Yolov5n

“…Loss functions in object detection models encompass various components, including confidence loss, bounding box coordinate loss, category loss, object loss, and segmentation loss for occlusion detection [28]. YOLOv4 [29] and YOLOv5 [30] build upon this foundation by incorporating components such as confidence loss, bounding box coordinate loss, category loss, Landmark loss, and Focal Loss, among others. The specific implementations of YOLOv4 and YOLOv5 may exhibit subtle differences in their loss functions, depending on specific implementation details and the libraries utilized.…”