License Plate Detection Based on Convolutional Neural Network and Visual Feature

In this paper, in order to solve the problem of license plate detection in license plate recognition, we propose a detection algorithm based on convolutional neural network and visual feature. First, we generate a certain number of candidate bounding box by means of artificial feature extraction. Then, the bounding box generated is used as input to the cascaded convolutional neural network for further verification and regression. After a series of experiments, our method has achieved good results both on accuracy and speed.