Smart trailer system with driver behaviour recognition and 360-degree surround view camera

Driver safety has always been an essential subject for humanity, and technology has improved continuously in the twenty century. New technologies for driver safety have started to develop in-car technologies. Even though the technologies are increasing and significant motor-vehicle fatalities are decreasing on motorways, driver safety still plays a big part in the automotive industry. This study intends to provide a new dataset and an effective method for driver behavior recognition based on driver behavior with a machine learning model named YOLOV5. Furthermore, a 360-degree blank spot surround-view camera system was also integrated into the study to increase driver safety. Due to advanced technology giving a chance to imply a better plan, using these systems may prevent traffic accidents much more rather than not using them at all. The image recognition model used in this study is YOLOV5 (You Only Look Once Version 5). This model is implemented the driver image data provided from KOE (Koluman Otomotiv Endüstri). Different parameters are used to determine the results. The selected metrics are mean averaged precision, precision, recall, classification, and essential other metrics. With the evaluation of the object behavior model's accuracies, this paper aims to determine the effectiveness of driver behavior recognition using the YOLOV5 model. Image processing methods used in this study are from OpenCV libraries. This library function is implemented to the taken image from fisheye lens cameras, and calibration methods are used for better results. Key words: Machine Learning, Image Processing, Driver Behavior Recognition, Surrond View, Artificial Intelligence