Traffic sign detection and problems in the field of computer vision

Abstract

Object detection is a typical task of computer vision. This paper presents some results of implementation of the traffic sign recognition. We use R-CNN for traffic sign detection system. We focus on speed limit superclasses of traffic sign. R-CNN deep learning detector is a simple and suitable model for the traffic sign recognition. This approach combines multiple low-level image features with high-level context from object detectors and scene classifiers. Despite the existing advances in computer vision, the article considers the problems that exist and which need to be solved in the future in the field of computer vision system design.