Deep Learning Technique to Analyze and Perceive Traffic Sign in the Intelligent Transport System

Abstract

Automobiles have evolved into a necessary part of our daily lives as a result of the rapid advancement of technology. One of them is advanced driver assistance systems (ADASs). The development of (ADAS) depends heavily on research into the automatic traffic sign detection and recognition (TSDR) system. Traffic sign detection is very important during the process of driving. Studies on vision-based TSDR have drawn a lot of attention from the research community, which is primarily motivated by three things: categorization, tracking, and detection. The detection of the traffic signs and taking precautionary majors is necessary while driving. This paper proposes the framework to detect the traffic signal and classify them into the different category. So, the driver will understand the severity of the signal. The proposed work consists of the modules as road sign detection, classification, and recognition. To identify the image, the color space conversion and color segmentation are applied. After the segmentation, the image is high lightened, and image is normalized and classified. For classify the different signals, neural networks are used. The proposed system is able to identify the cautionary sign, mandatory and informative signs. This paper evaluated some of the traffic signs such as no entry, stop, give way sign, and speed limit sign.