Research On Vehicle Detection Algorithm And Application Based On Attention Mechanism

With the development of artificial intelligence technology,the application of intelligent transportation is more and more extensive.The use of computer vision technology to detect and identify traffic images can be used in traffic detection,automatic driving,vehicle speed measurement and other scenes.Aiming at the problem that current vehicle detection methods are insufficient in extracting features of vehicles,leading to inaccurate vehicle positioning and classification,this paper studies the vehicle detection method based on attention mechanism and the vehicle detection method based on feature fusion.Compared with other methods,it has higher recognition accuracy.The main research contents are as follows:Firstly,the model detection algorithm based on the attention mechanism is proposed,and the model detection is carried out by using Center Net,the key point estimation network.By introducing the attention mechanism,the weight learning is carried out on the features of different channels and different positions,so that the feature extraction network can learn vehicle features targeted;in addition,the connection between different layers of the feature extraction network is added to fuse features of different scales.The KITTI vehicle data set and BIT-Vehicle data set are expanded by means of data enhancement,and the detection accuracy of 94.6% and 95.5% are achieved respectively through experiments.Then the vess’ shicle detection algorithm based on multi-scale feature fusion is proposed.Resnet-50 is used for feature extraction.Through hierarchical fusion of vehicle features of the same scale,different receptive fields are obtained to perceive local information and global information.Different feature fusion methods in depth and width are used to fuse the features of vehicle models,and the detection accuracy of 95.9% is achieved on the BIT-Vehicle dataset.The importance of hierarchical feature fusion for vehicle detection task is verified through experiments.Finally,an intelligent vehicle detection prototype system is designed by using the vehicle detection algorithm in this paper.The user can detect the position and type of the vehicle in the image by calling the algorithm interface through the interface,providing the basic data of vehicle type detection for intelligent transportation.