WINDSHIELD, MOBILE PHONE, AND SEAT BELT DETECTION USING YOLOV11 FOR DRIVER MONITORING SYSTEMS

Road accidents remain a major public safety concern in Algeria, with 26,272 accidents, 3,740 deaths, and 35,556 injuries recorded in 2024. Over 90% of these accidents are caused by human error, primarily due to behaviors such as the use of mobile phones while driving and failure to wear seat belts. In response to this critical issue, this paper introduces an advanced detection system to enhance road safety through real-time monitoring of mobile phone usage and seat belt compliance among vehicle occupants. The proposed system uses the state-of-the-art YOLOv11 architecture and consists of three interconnected components: seat belt detection, mobile phone usage detection, and windshield detection. YOLOv11 incorporates key innovations, including C2PSA blocks that enhance spatial awareness and C3k2 modules with smaller kernel sizes, enabling better identification of small and partially occluded objects. The system was developed using a dataset from Roboflow comprising 1,116 images, which was expanded to 2,139 images through augmentation techniques such as flipping, brightness adjustment, and rotation. Training was conducted in Google Colab's graphics processing unit environment with 100 epochs and a batch size of 16. Performance evaluation shows the system’s reliability, achieving a mean average precision of 94.2%, with detection accuracies of 96.5% for seat belts, 99.5% for windshields, and 86.5% for mobile phone usage. These results indicate that the proposed system is a promising tool for monitoring driver behavior and supporting the enforcement of safety regulations, thereby contributing to efforts to reduce accident rates in Algeria.