LPRNet: License Plate Recognition via Deep Neural Networks
This addresses the problem of efficient and accurate license plate recognition for embedded systems, particularly for challenging Chinese plates, though it is incremental as it builds on existing deep learning approaches.
The paper tackles license plate recognition by proposing LPRNet, an end-to-end deep neural network method that avoids character segmentation, achieving up to 95% accuracy on Chinese plates with real-time speeds of 3 ms/plate on GPU and 1.3 ms/plate on CPU.
This paper proposes LPRNet - end-to-end method for Automatic License Plate Recognition without preliminary character segmentation. Our approach is inspired by recent breakthroughs in Deep Neural Networks, and works in real-time with recognition accuracy up to 95% for Chinese license plates: 3 ms/plate on nVIDIA GeForce GTX 1080 and 1.3 ms/plate on Intel Core i7-6700K CPU. LPRNet consists of the lightweight Convolutional Neural Network, so it can be trained in end-to-end way. To the best of our knowledge, LPRNet is the first real-time License Plate Recognition system that does not use RNNs. As a result, the LPRNet algorithm may be used to create embedded solutions for LPR that feature high level accuracy even on challenging Chinese license plates.