An Efficient Traffic Sign Classification and Recognition with Deep Convolutional Neural Networks

Abstract

Driver safety assisted driving, and autonomous driving all depend on automated traffic sign recognition. The
most popular deep learning method for recognizing traffic signs is convolutional neural networks. (CNNs).
This paper presents a useful technique for automatically detecting images of traffic signs. When employing the
two common traffic sign photo datasets GTSRB, our method makes use of our CNN model architecture and
performs the best. It aids the driver in safely operating the motor vehicle. The amount of time and effort drivers
spend manually evaluating and recognizing traffic signs is excessive. This work provides an autonomous traffic
sign identification using a convolutional neural network. Our work here introduces a unique CNN architecture
with an Adam optimizer and a batch size of 128 to improve the efficiency of traffic sign recognition. Results
based on a complex network were more accurately produced by a convolutional neural network (CNN). With
99.81% precision and a minimum of losses, our system learns from the GTSRB dataset, which contains 43 traffic
classes, to identify the correct class of an anonymous traffic sign. The results, however, are better than those of
the prior research, which examined this approach’s accuracy and efficiency in recognizing traffic signs despite
poor weather and hazy image circumstances.