Multi-task convolutional neural network system for license plate recognition

License plate recognition is an active research field as demands sharply increase with the development of Intelligent Transportation System (ITS). However, since the license plate recognition(LPR) is sensitive to the conditions of the surrounding environment such as a complicated background in the image, viewing angle and illumination change, it is still difficult to correctly recognize letters and digits on LPR. This study applies Deep Convolutional Neural Network (DCNN) to the license plate recognition. The DCNN is a method of which the performance has recently been proven to have an excellent generalization error rate in the field of image recognition. The proposed layer structure of the DCNN used in this study consists of a combination of a layer for judging the existence of a license plate and a layer for recognizing digits and characters. This learning method is based on Multi- Task Learning (MTL). Through experiments using real images, this study shows that this layer structure classifies digits and characters more accurately than the DCNN using a conventional layer does. We also use artificial images generated directly for training model.