An effective automatic traffic sign classification and recognition deep convolutional networks

Abstract Automatic traffic sign recognition is essential for autonomous driving, assisted driving, and driver safety. Convolutional neural networks (CNNs) are the most widely used deep learning algorithm for traffic signs recognition. This research presents an effective technique for automatically r...
Ausführliche Beschreibung

Abstract Automatic traffic sign recognition is essential for autonomous driving, assisted driving, and driver safety. Convolutional neural networks (CNNs) are the most widely used deep learning algorithm for traffic signs recognition. This research presents an effective technique for automatically recognizing traffic signs images. This technique mainly uses four GTSRB, GTSDB, BTSC, and TSRD standard traffic signs images datasets of different traffic sign images and provides the best result using our CNN model architecture. It helps to assists the driver in driving the motor vehicle safely. Drivers devote too much attention and effort to recognizing traffic signs by manually analyzing and recognizing their aspects. This study presents an automatic traffic sign recognition system to minimize motor vehicle accidents using representation to identify signs-this task utilizing a deep convolutional neural network. Here, our work presents a novel CNN architecture (.001) with Adam optimizer, batch size 128, and multi-interconnect layers to improve the performance of traffic signs detection. A Convolutional Neural Network (CNN) achieved more accuracy, results based on a complex network. Our model learns from the GTSDB dataset, which contains 43 traffic classes, and uses this information to predict the proper class of an anonymous traffic sign with 99.81% accuracy and minimum losses. Contradictory, the result is improved than the earlier study, which examined 98.20% accuracy that the approach can still detect traffic signs with extreme weather conditions and blur image conditions. Ausführliche Beschreibung