Abstract:
Nowadays road accident is a very major issue for human death. Among road accidents, bike accidents are the biggest problem in our country. Every year, the number of deaths is increasing from the previous year. Bike accidents cause serious injuries to people because off-bikers do not use helmets. One of a motorcyclist's essential protective devices is a helmet. After there are laws and fines in our country about the use of helmets, bikers do not follow them. Solving this problem manually is a matter of a lot of money and time. So a system needs to be made which can automatically classify those who wear helmets and those who don't. There is already a system in place that wealthy nations have created that Bangladesh cannot afford. As a result, I made the decision to develop a system that will help in the authority's classification of helmet detection and number plate recognition. An image processing and convolutional neural network system are used in this case to identify the motorcyclists who are breaking the helmet rules. The system consists of motorcycle detection, categorization of wearing a helmet vs not wearing one, and motorcycle license plate identification. Using the YOLOv7 function, the motorcycles are detected. Once the motorcycle has been identified using a convolutional neural network, it is decided whether or not the rider is wearing a helmet. When a rider without a helmet is recognized, Open CV Tesseract OCR is used to find the motorcycle's license plate. In the future, I will work on automatic case file systems that violate the helmet law and will increase the dataset to get higher accuracy.
