Optimizing Traffic Safety: YOLOv11 and DeepSORT for Speed Detection and Safe Distance Monitoring

Considering the high accident rate, improving traffic safety is essential. Because traditional monitoring systems still rely on human management, it is more challenging to identify vehicles that are speeding and failing to keep a safe distance from one another. This study proposes the integration of YOLOv11 and DeepSORT to detect vehicles exceeding a certain speed limit and not maintaining a distance between vehicles in real traffic. Using Area Traffic Control System data, our model achieved 81% accuracy on the classification task and 87.3% mAP, 83% precision, 80.1% recall, and 81.5% F1-Score on the detection task—the model evaluation for the classification task from the modeling results, with 92% accuracy. Compared to studies that only detect, track, measure speed, and count vehicles, the system built has an improvisation. Despite the limited data and pixelated images, the system shows good results.