Enhancing Vehicle License Plate Recognition in Low-Light Conditions Using YOLOv5 and Advanced Pre-processing Techniques

License Plate Recognition (LPR) is a real-time system that converts optical images of vehicle license plates into digital data. LPR, which is widely employed in transportation systems, also benefits healthcare by enhancing parking management and optimizing ambulance routing. The implementation of a vehicle license plate recognition system for a low-light environment is discussed in this project. The main goal is to increase the detection and recognition accuracy in low light. To achieve robustness, the YOLOv5 object detection technique was trained on a dataset with a range of lighting conditions using the YOLOv5 object detection technique. Pre-processing techniques such as image scaling, grayscale conversion and contrast enhancement were used to improve the image clarity. Before Optical Character Recognition (OCR), the images were enhanced by advanced techniques such as de-noising and homomorphic filtering. The YOLOv5 was 96% accurate in the detection phase, detecting 96 out of 100 license plates correctly. The system then achieved an overall OCR accuracy of 88%. Future recommendations include using infrared or thermal imaging in low light conditions, better de-blurring algorithms, and optimization for high-speed vehicle identification and real-time processing. The changes that these bring intend to make the system more robust and versatile so that the license plate identification is dependable in any demanding setting.