Traffic accidents are a serious problem worldwide, resulting in thousands of lives lost and millions injured every year. One of the greatest causes of accidents is driver negligence in maintaining a safe distance between vehicles. This research was performed to develop an advanced driver assistance system (ADAS) system capable of real-time measurement of the distances between vehicles and objects or from vehicles in front on a mobile application that can provide warnings when such distances become too small. This system is a simplified version of expensive sensors or lidars, which are difficult to integrate into vehicles. The system uses the YOLOv5 model to detect and identify vehicles using a smartphone camera. The YOLOv5 model training resulted in a mean average precision (mAP) value of 90%, indicating high object detection capability. The system was integrated into a mobile application to enable real-time vehicle detection. Tests showed that the system could detect and identify vehicles as well as predict the distance between vehicles with an accuracy of 94.08%. This research showed that the system can replace both sensors and lidar for measuring the distances to objects in front of a vehicle in real-time using only a smartphone.
DrivEye: A Computer Vision-Based Object Detection System to Measure the Distance Between Vehicles
12.09.2024
1042164 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
The Optimal Distance Measure for Object Detection
British Library Conference Proceedings | 2003
|The optimal distance measure for object detection
IEEE | 2003
|Vision Based Object Distance Estimation
British Library Conference Proceedings | 2017
|