Vision-based automatic warning system to prevent dangerous and illegal vehicle overtaking

Abstract

The purpose of this research is to implement a system that can be a help for drivers to drive vehicles safely and reducing accidents. Most of the accidents have occurred when overtaking. The driver has to consider the distance between two vehicles and the speed of the front vehicle. If there is another vehicle moving the same or opposite direction, it must be considered as well. Also, traffic signs, solid line, and a broken line must be considered. Reducing the errors which can be happened by the driver when deciding all these stuff in instinct is the goal of the “Automated Vehicle Overtaking System”. This research has been used three pipelines, such as lane line detection, traffic sign detection, and vehicle detection. The input video stream from the front view camera of the vehicle is transferred through the above pipelines. The system gives the final output frame with confirmation that ‘is it safe to overtake now’. The driver can recognize the danger by alarm sound from the system. This system has a Python Application Programming Interface (API). Using that API, any device or application can access this system. Finally, the main goal of this research is to implement a system that can predict dangers in overtaking and saves the lives of passengers and vehicles.