Enhancing Road Safety: Inter-Vehicle Distance Estimation Using Fine-Tuned YOLOv9

The estimation of inter-vehicle distance is an important constituent part of current as well as future-generation Advanced Driver Assistance Systems (ADASs) focused on road safety improvement. YOLOv5, YOLOv7, YOLOv8, and YOLOv9, with their state-of-the-art versions, and Mask R-CNN are applied for detection purposes along with the computation of inter-vehicle distance. The models estimate the distance by making use of the real-time self-developed data provided by video feeds and images using vehicle bounding boxes and center point calculations. Amongst those, our proposed fine-tuned YOLOv9 attained the best result with a Mean Average Precision (mAP) of 0.910, a precision of 0.8903, and a recall of 0.8189. Results from this work indicate that such systems based on YOLO can be effectively integrated into ADAS’s for safer driving by providing the most accurate real-time inter-vehicle distance estimations using the model YOLOv9. Future advancements in autonomous driving systems stemming from this work depend on the reliability of distance measurements under various conditions.