<![CDATA[The application of CCTV cameras for traffic surveillance and monitoring is one effective solution to address urban traffic problems, as the number of vehicles that continue to increase rapidly but the area of the road remains the same will cause congestion. However, the problem in traffic surveillance and monitoring is not just focusing on vehicle detection based on category inference on video sequence data sourced from CCTV cameras alone, another important, challenging task is to combine calculations, classification and tracking of different vehicle movements in urban traffic control systems. The study expanded on previous research by breaking down the problem into different sub-tasks using the YOLOv4 approach combined with the Deep Sort algorithm for the detection and tracking of objects directly on CCTV footage of vehicle activity on the city's three-stop highway. Based on the results of YOLOv4 testing resulted in a detection accuracy rate with mAP of 87.98% where the combination of YOLOv4 with the Deep Sort algorithm can detect, track and calculate 13 types of vehicles.]]>
Copyrights © 2021
