PENERAPAN YOLOV5 DAN SORT DALAM DETEKSI KENDARAAN PADA PERSIMPANGAN BERSINYAL UNTUK PENYESUAIAN WAKTU LAMPU LALU LINTAS

Abstract

The problems with traffic in many big cities is congestion. Detection and tracking of vehicles in traffic light queues is an important aspect of an efficient urban transport system. This research introduces a combined model that combines the YOLO (You Only Look Once) version 5 algorithm or commonly called YOLOv5, a high-speed object detection approach, with the SORT (Simple Online and Realtime Tracking) algorithm to efficiently detect and track vehicles. The method was tested using a dataset of real traffic recordings and the results showed excellent performance in detecting and tracking vehicles, with sufficient accuracy and speed for urban traffic applications. Using this approach, vehicle detection and tracking systems can improve traffic safety and optimize vehicle flow at road intersections.