Vehicle tracking algorithm based on deep learning

Abstract

Vehicle detection is one of the most important detection targets in target detection. overcoming this problem is of great significance in the field of traffic detection and automatic driving. In this paper, based on yolov4, firstly, the slice module is added to the input, and the silce module slices the image into the backbone to get the sampling feature map without information loss. Then, the SPP module is added to the neck of neck module, and then the Kitti dataset is trained, detected and tested. Experiments show that adding slice module in the input circuit and spp module in the neck can effectively improve the accuracy of vehicle detection Map@0.5 3.4%.