The subject of the work focuses on the field of computer vision, more specifically on the detection and recognition of license plates. As part of this work, an algorithm was created that identifies license plates from photos, and then interprets the registration numbers, converting images of license plates into readable strings of characters. The YOLOv4 neural network model was used for the detection process. The work contains a detailed explanation of the selected techniques, described technologies used along with implementation issues. The software was subjected to detailed tests on various cars. Experiments were conducted at two different times of the day, which allowed for the assessment of the algorithm’s performance in different lighting conditions. Additionally, the tests included various distances and angles at which the license plate was visible to the camera, which allowed for the assessment of the algorithm’s ability to cope with various real-world scenarios.
Detecting Plates with diffrent light scenario
Accuraccy tested on various distance and angels