Traffic Sign Detection using Deep Learning Techniques in Autonomous Vehicles

Abstract

Autonomous vehicle is an emerging topic for both researchers and the automobile industry as companies are still struggling to make fully functional autonomous vehicles. Driving a safe vehicle in a real world depends on different conditions, such as distance from other vehicles, pedestrians, animals, speed-breakers, traffic signals and other unpredictable dynamic environments. Autonomous vehicle can decrease vehicle crashes because software installed in the vehicle instructs the control system of the autonomous vehicle rather than human, and Software makes less error compare to human beings. Automated Traffic Sign Detection and Recognition (ATSDR) is an important task for a safe driving by an autonomous vehicle. Many researchers have used various deep learning-based models for in real-time ATSDR. Here in the present review, we have studied various deep learning models used for in real-time ATSDR. Our study suggested that YOLO and SSD can detect the traffic sign in real time and are superior models for ATSDR as compared to other deep learning methods as CNN, R-CNN, Fast R-CNN and Faster RCNN.