A CNN-Based Approach for Automatic License Plate Recognition in the Wild

Meng Dong, Dongliang He, Chong Luo, Dong Liu and Wenjun Zeng

Abstract

In this paper, we address automatic license plate recognition (ALPR) in the wild. Such an ALPR system takes an arbitrary image as input and outputs the recognized license plate numbers. In the detection stage, we adopt a cascade structure comprising of a fast region proposal network and a R-CNN network. The R-CNN network not only eliminates false alarms but also regresses corner positions for each detected plate. This allows us to estimate an affine transformation matrix to rectify the extracted plates. In the recognition stage, we propose an innovative structure composed of parallel spatial transform networks and shared-weight recognizers. The system is trained and evaluated on a Chinese license plate dataset with over 18K images. Results show that our detector performs better than faster R-CNN (VGG) which is 1.5x slower in testing and 57x larger in model size. The recognizer is also significantly better than existing solutions, reducing 57.

Session

Posters

Files

Paper (PDF)
Supplementary (PDF)

DOI

10.5244/C.31.175
https://dx.doi.org/10.5244/C.31.175

Citation

Bibtex

            @inproceedings{BMVC2017_175,
                title={A CNN-Based Approach for Automatic License Plate Recognition in the Wild},
                author={Meng Dong, Dongliang He, Chong Luo, Dong Liu and Wenjun Zeng},
                year={2017},
                month={September},
                pages={175.1-175.12},
                articleno={175},
                numpages={12},
                booktitle={Proceedings of the British Machine Vision Conference (BMVC)},
                publisher={BMVA Press},
                editor={Tae-Kyun Kim, Stefanos Zafeiriou, Gabriel Brostow and Krystian Mikolajczyk},
                doi={10.5244/C.31.175},
                isbn={1-901725-60-X},
                url={https://dx.doi.org/10.5244/C.31.175}
            }