Physical Adversarial Example in Car Plate Recognition Systems Based on SA Algorithm

Currently, deep learning technology has become a crucial branch of artificial intelligence, particularly in consumer electronics, where it brings numerous innovative applications through classification and object detection techniques. However, research shows that deep neural networks are susceptible to adversarial example. This brings a series of security issues for related applications. Moreover, adversarial example can also directly exist in the real world and affect models. In this study, we investigate the impact of physical adversarial example on car plate recognition systems and explore the future development and challenges related to this topic. Our simulation results show that adding specific patterns to the surroundings of a car can generate a physical adversarial example, causing the car plate recognition system to fail.