Vehicle Detection in Traffic Camera Perspective with Improved YOLOv5s
LIN Haifeng
LIU Dapeng
CAI Hui
Abstract：Aiming at the problems of dense vehicles in the current traffic camera perspective,a vehicle detection algorithm based on improved YOLOv5s is proposed from the traffic camera perspective.Firstly,in order to enhance the feature extraction abili-ty of the network,the CA attention mechanism and the BotNet network are introduced to improve the recognition rate of vehicles.Then,in order to make the target box regression loss of the network in the training phase converge faster and improve its positioning ability and reasoning performance,SIoU is used to replace the original frame regression loss function CIoU.Finally,in order to im-prove the detection ability of small targets,a shallower feature layer is added to the detection head,which changes from three scale detection to four scale detection to detect smaller targets.The experimental results show that the improved YOLOv5s improves the av-erage accuracy by 2.9%compared with the original algorithm,and the detection speed reaches 51 FPS,which has certain accuracy and real-time performance.
Keywords：YOLOv5CA attentionBotNetSIoUobject detection
Publication Date：2025-05-20
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：7( 1327-1332,1398 )