A Deep Learning Model Of Traffic Signs Detection

Abstract

To pursue the ideal of a safe high-tech society in a time when traffic accidents are frequent, the traffic signs detection system becomes one of the necessary topics for the future. The ultimate goal of this research is to be able to identify and classify the types of traffic signs in a panoramic image. To accomplish this goal, the research is based on Convolutional Neural Network (CNN) and Mask RCNN available to modify and develop traffic sign detection model in the panoramic images. Data augmentation and normalization of the images will also be applied to assist in classifying better even when old traffic signs are degraded, and considerably minimizes the rates of discovering the extra boxes. The final model got approximately 94.5% of the correct signal recognition rate, the classification rate of those signs discovered was approximately 99.41% and the rate of false signs was only around 0.11. The orthodox Mask RCNN model is researched and implemented at: https://github.com/facebookresearch/detectron2.