Road sign detection and recognition system based on multi-layers convolutional neural network model trained with German Traffic Sign Recognition Benchmark
Siwei Fan · 2021 4th World Conference on Mechanical Engineering and Intelligent Manufacturing (WCMEIM) · 2021
Traffic sign detection and recognition play a vital role in the development of self-driving systems. It helps to increase the safety of drivers during the driving process. However, its accuracy will be sometimes affected by certain environmental factors and the accuracy needs to be improved. This paper proposes a detection and recognition system with a 8 layers convolutional neural network that acquires different kinds of features by training with different kinds of traffic signs. The training made an outstanding contribution to our detection and recognition system, allowing our system to be more comprehensive and precise compared with most current systems, which solved the problems of limited accuracy and performance under the effects of environmental factors.