Deep Learning Traffic Sign Recognition in Autonomous Vehicle

In this paper, a deep learning method is used to make a system for traffic sign recognition. You Only Look Once (YOLOv3) is used as it has a quick response in terms of real-time data reliability followed by high accuracy and robust performance. This study applies image preprocessing for better decision making for the recognition system in a different environment which includes lighting and weather. This is to ensure that the approach used is safe to be installed in autonomous vehicles. A comparison of images trained and tested will be demonstrated. The accuracy reach up to 100% and time to recognize traffic sign in image is in 36.907457 seconds. An analysis is done to ensure the error rate is reduced as training is done in a longer $period$.