Small Road Object Detection Algorithm based on Improved YOLOv8
In order to solve the problem that the original YOLOv8 is prone to false detection and missing detection of small objects in traffic road scenes,the paper presents a detection method based on the improved YOLOv8 network model.Firstly,the downsampling mechanism(ADown)is used to replace the traditional convolution in the trunk to expand the receptive field of the model.Secondly,the CBAM attention mechanism was introduced to reduce the missing detection rate of small objects.Finally,a detection head is added to improve the detection accuracy for small targets.Experiments show that the recall R,mAP@0.5 and mAP@0.95 of the improved algorithm on the KITTI dataset are increased by 0.1％,1.7％and 3.1％,respectively,while the number of parameters decreases by 11％,and that the indicators for the detection effect of small targets are all improved.These results prove that the improved algorithm is effective in the detection of small road objects.
万方数据
维普
