Research on Road Vehicle Detection based on improved Yolov5s

Vehicle target detection is a very important technology in the field of autonomous driving, This paper presents a road vehicle detection method based on improved Yolov5s. First, SiLU activation function is used to replace the original activation function in Yolov5s backbone network, the number of parameters in the network structure is reduced and the problem of gradient disappearance is solved. CBAM (Convolutional Block Attention Module) attention mechanism module has been added to Backbone, enhanced feature extraction of small and medium target objects in the network. The results show that compared with the original Yolov5s model, after the improvement, the model size is reduced by 0. 4MB, map@0.5 is increased by 4.9% to 92.3%, precision increased 2.6% to 94.5%, the improved Yolov5s model is able to detect vehicles ahead more efficiently and accurately.