INTELLIGENT TRANSPORT SYSTEM FOR SELF DRIVING CARS

Abstract

Traffic light recognition plays an important role in intelligent transportation systems and modern autonomous driving technologies. Vehicles operating in urban environments must be able to correctly interpret traffic signals in order to make safe and timely driving decisions. However, recognizing traffic lights from road images is a challenging task due to several factors such as changes in lighting conditions, weather effects like rain or fog, partial occlusions, and the small size of traffic lights in large scene images. These challenges often reduce the accuracy of traditional detection methods that rely on basic image processing techniques. To address these issues, this research presents a real-time traffic light recognition system based on deep learning. The proposed approach combines the YOLOv3 object detection model with a lightweight Convolutional Neural Network (CNN) classifier known as TL-Net. In this framework, YOLOv3 is first used to identify and locate traffic lights within an input image by detecting the region of interest. The detected regions are then passed to the TL-Net classifier, which determines the state of the traffic light as red, yellow, or green.The performance of the proposed system is evaluated using well-known public datasets such as the DriveU Traffic Light Dataset (DTLD) and the LISA Traffic Light Dataset. Experimental results show that the model can accurately detect and classify traffic lights even in complex traffic scenes. Compared with conventional image processing methods, the proposed deep learning approach provides higher accuracy and better real-time performance, making it suitable for applications in intelligent transportation systems and autonomous vehicles.