Research On Traffic Sign Detection And Recognition Algorithm

Road sign recognition is an important part of the intelligent transportation system.In the real environment,the complex external environment causes problems such as excessive or insufficient lighting,obstacles,and unstable shooting.These have a great impact on the detection and recognition of road signs,and the accuracy and real-time nature of traffic sign detection and recognition It is the actual problem currently facing.This paper uses deep learning technology to research traffic sign detection and recognition algorithms.The main contents of this paper are as follows:(1)In the stage of traffic sign detection,the most commonly used single-stage detection algorithms for target detection are studied: YOLO series method and SSD method.Analyze the advantages and disadvantages of these two types of models.Then,a comparative experiment was carried out on the German traffic sign data set to compare the performance of the three algorithms YOLO v2,YOLO v3 and SSD.Through the comparison,YOLO v3 was selected as the detection model.(2)In the recognition stage,an improved multi-layer feature fusion LeNet-5recognition algorithm is proposed to recognize and classify road signs.The improved CNN network optimizes the number and size of convolution kernels,makes the model more sensitive to the characteristics of the target object,adds a batch normalization BN layer(input normalization layer),and changes the activation function to a non-linear RELU function to avoid The appearance of the gradient disappearance phenomenon is reduced,the calculation complexity of the activation function is reduced,and the Dropout strategy is adopted to improve the robustness of the model and effectively avoid over-fitting.(3)The GTSRB data set is expanded through the data enhancement method,which balances the distribution of data set types,increases the amount of training data,and improves the generalization ability of the data set.On this basis,it is based on YOLO v3 and improved multi-layer LeNet-5 traffic sign detection and recognition.The experimental verification results show that the algorithm proposed in this paper has good accuracy and real-time performance in the detection and recognition of traffic signs in complex environments.