TRAFFIC LIGHT DETECTION AND CLASSIFICATION USING RESNET50

Abstract

Improving road safety and optimising traffic management are two of the most 
important functions of Intelligent Transportation Systems (ITS). The quick and 
precise identification and categorisation of traffic signal conditions is crucial to the 
efficiency of these systems. By introducing a new approach to real-time traffic signal 
detection, this study adds to the growing body of work in ITS-related computer vision 
and machine learning. By using the ResNet (Residual Networks) architecture, our 
method tackles the complex issues caused by many environmental factors, such as 
bad weather, different lighting, and obstructions. Our system is able to detect 
complicated patterns and characteristics thanks to ResNet's deep learning capabilities, 
demonstrating its flexibility to real-world traffic circumstances. The importance of 
precise traffic light detection and categorisation is shown by the research, which goes 
beyond the technical details. Optimal traffic flow, reduced congestion, and improved 
road safety are all goals of these procedures, which are more than just technical 
requirements in today's ever-changing urban landscapes. With the goal of offering 
workable answers to the problems encountered by actual ITS systems, this study adds 
to the larger conversation on computer vision and machine learning. Our work aims to 
drive progress in the area by highlighting the significance of precise traffic light 
identification. This might lead to smart traffic control systems that improve urban 
safety and mobility.