I have worked on many OCR architectures. I have even trained ANPR for 24 different European languages and ANPR for India. Basically there is no pre-trained recognizer for OCR of license plates. You need to train a model for that. You can choose multiple OCR architectures and see which one is the best for your application. For training the OCR model, you need to first prepare a dataset of license plate images. You can scrape online images, crop them from CCTV cameras on roads, label them manually or from OCR service like amazon or google and create a good quality dataset.