An Effective Method for Yemeni License Plate Recognition Based on Deep Neural Networks

As the existing publicly license plate recognition (LPR) datasets available for training are restricted and almost non-existent in some countries, for example, some developing countries. In this paper, first we present the first Yemeni License Plate dataset (Y-LPR dataset) includes vehicles and license plate images for Yemeni license plate detection and recognition. Second, we propose a new LPR method for license plate detection and Recognition. It consists of two key stages: First, License plate detection from images based on the latest state-of-the-art deep learning-based detector which is YOLOv5. Second, Yemeni Character and number recognition based on the CRNN model. Experimental results show that our method is effective in detecting and recognizing license plates.