Automated Non-Helmet Rider Detection using YOLO v7 and OCR for Enhanced Traffic Monitoring

Motorcycle accident victims who have head injuries are more likely to die or become disabled. Head injuries can occur more rarely and are less severe when a helmet is worn, research has indicated. Helmet use can prevent facial injuries, which can be particularly debilitating and disfiguring. Despite the benefits of wearing helmets and the evidence supporting their effectiveness, many motorcyclists still choose not to wear them due to a variety of reasons such as discomfort, inconvenience, and lack of awareness. This poses a serious safety risk not only to the rider but also to other road users. Fines must be enforced in order to ensure that the public complies with the helmet-wearing laws established in the relevant country’s motor vehicle act. To achieve this, a deep learning model for Automatic non-helmet rider detection and number plate recognition is posed in this paper wholly based on the efficiency of YOLOv7 single-shot detector algorithm. YOLOv7 is used for object detection and to classify helmet, non-helmet riders. After the classification, the images with non-helmet riders are treated with OCR also known as Optical Character Recognition, which is a technology where it allows the system to recognize text within images and extract the characters from the License plate. Hence the respected authorities can impose fines or take necessary actions for the violations of the traffic rule.