Nighttime Rainy Season Traffic Analysis: Vehicle Detection, Tracking, and Counting with YOLOv8 and DeepSORT

Abstract

This research focuses on developing a reliable computer vision system for accurately tracking traffic density in India during the rainy season. The system uses deep learning-based techniques to handle the difficulties associated with vehicle detection and tracking. The three modules are vehicle detection, tracking, and vehicle counting. Vehicles are initially identified using the YOLOv8 algorithm, a state-of-the-art deep learning detector. Subsequently, the DeepSORT algorithm is utilized for multi-object tracking to ensure accurate and robust tracking of various objects, including cars, buses, trucks, bikes, and pedestrians. The importance of accurate vehicle counting and speed measurement is emphasized, especially during bad weather. An independently compiled dataset of Indian rainy conditions is used to assess the proposed computer vision system. The outcomes demonstrate the system's capability to accurately identify, track, count, and estimate the speeds of vehicles. These features offer insightful information for traffic analysis, including flow monitoring, congestion detection, and other associated traffic challenges. This study makes a contribution to the field of computer vision-based traffic monitoring and offers potential applications in transportation management systems under challenging weather conditions.