Vehicle image recognition model establishment and recognition method

The invention discloses a method for establishing a vehicle image recognition model, which comprises the following steps: using a common public vehicle image data set, augmenting and preprocessing the vehicle image data set, constructing and training a multi-branch convolutional neural network to obtain the multi-branch convolutional neural network for vehicle recognition, and utilizing the multi-branch convolutional neural network obtained by training to establish a vehicle image recognition model. And converting the structure into a single-branch convolutional neural network through re-parameterization, extracting features of the vehicle image, identifying and classifying the features, and finally obtaining a vehicle type. According to the method, a multi-branch convolutional neural network is improved, a spatial pyramid is used, and pooling results are spliced, so that the network can extract features of different scales, and extraction and fusion of vehicle image features are realized; according to the ve