Research on Vehicle Detection and Vehicle Type Recognition under Cloud Computer Vision

Abstract

Cloud computer vision technology has a broad application prospect in transportation system, which provides more intuitive and convenient analysis means and advanced technology for the realization of vehicle assisted driving. In this paper, the video stream of traffic image captured by fixed camera is taken as the research object, and the real-time extraction of traffic parameters is taken as the research purpose, and the detection algorithm of traffic flow parameters is studied and improved. In this paper, the application of automatic encoder in vehicle recognition is studied, and the performance of automatic encoder features in vehicle recognition is analyzed. On the basis of research, the convolution neural network and sparse automatic encoder network are fused to extract convolution features of vehicles, and the deep convolution automatic encoder network is trained by greedy training method layer by layer to realize vehicle classification. The results show that the vehicle detection network can achieve high detection accuracy in different environments, and the vehicle classification network can achieve good classification results for vehicles in traffic videos.