Traffic sign recognition system using CNN and Keras

Abstract

In this paper, we propose the best approach for a Traffic sign recognition system with a high accuracy rate and less computing time. This process is done with help of CNN and Keras. In fully automatic driving cars, it is difficult to recognize the traffic signs with less computing time and a high accuracy rate. So, to solve this problem, first, we are exploring the sample traffic sign dataset, next images are sorted and their labels are set into a list and those lists are converted into NumPy arrays for feeding to the model. Secondly, the CNN model is built to classify the images into their respective categories, this is the best approach for image classification. After building the model, the model is trained, validated, and tested using the test dataset. Finally, the graphical user interface is built for traffic sign recognition using Tkinter.