Key points are not available for this paper at this time.
Over the past few years, breakthroughs in deep learning applied to computer vision has driven the development of novel techniques and various solutions for automatic license plate recognition (ALPR). Albeit this technology has been successfully used in constrained conditions, in-the-wild scenarios have emerged as the new challenge for these systems. In-the-wild scenarios, also referred to as complex natural scenes or open scenarios, involve using regular video surveillance in various settings, including but not limited to drones and security robots. This survey is devoted to reviewing the state-of-the-art associated with in-the-wild ALPR. From a literature-based perspective, we (i) highlight specific approaches used for license plate (LP) detection, LP rectification, and LP recognition; (ii) present some of the most widely used datasets for benchmarking and summarized the results reported in the reviewed studies; and (iii) outline key research directions for advancing in-the-wild ALPR. This paper aims to provide practitioners, researchers, and experts with a comprehensive understanding of the current state of ALPR systems and the associated challenges, particularly in-the-wild scenarios.
Ismail et al. (Wed,) studied this question.