Traffic Sign board image Classification by using Deep Learning Techniques

Abstract

Abstract In this research paper, the comparison of various regional image enhancement methods is made with three-dimensional image quality statistics such as mean error (MSE), peak signal-to-noise ratio (PSNR), and image quality. (SSIM) Image quality. Image development activities can be carried out, especially in local regions. The criteria for choosing the best method are the characteristics of the mean square error (MSE) and cumulative mean square error (MSE) to noise ratio (PSNR), as well as the maximum measurement and image quality (SSIM). Classification of tasks from preliminary images using the pre-learning technique of deep convolutional neural network. Accuracy for testing and training image databases is 100%.