Computer Vision System for Speed Limit Traffic Sign Recognition

The paper is devoted to the development of a portable speed limit traffic sign recognition system widely available for general use. A shape-based approach has been applied to detect a speed limit sign. The sign recognition has been carried out with a convolutional neural network. A single board computer was used as a prototype of a hardware part of the proposed system. The hardware and software parts of the proposed system have been developed and tested. The traffic sign detection range is 25 meters in case of good lighting, the average recognition time is 5 seconds. The methods and techniques to increase system performance have been proposed.