Towards Automatic License Plate Recognition in Challenging Conditions

License plate recognition (LPR) is an integral part of the current intelligent systems that are developed to locate and identify various objects. Unfortunately, LPR a challenging task due factors, such as numerous shapes designs LPs, non-following standard LP templates, irregular outlines, angle variations, occlusion. These factors drastically influence appearance significantly challenge detection abilities state-of-the-art algorithms. However, recent rising trends in development machine learning algorithms have yielded encouraging solutions. This paper presents novel method address aforesaid issues. The proposed composed three distinct but interconnected steps. First, vehicle appears input image detected using Faster RCNN. Next, area located within by morphological operations. Finally, license accomplished deep network. Detailed simulations performed on PKU, AOLP, CCPD databases indicate our approach produces mean accuracy 99%, 96.0231%, 98.7000% databases.