Traffic sign detection and recognition based on convolutional neural networks

Traffic sign detection and recognition topic are one of the most popular topics of computer vision and image processing in recent years, as they play an important role in autonomous driving and traffic safety. This paper proposes a system that will detect and classify different types of traffic signs from images. This paper differs from other papers as it uses signs that are globally recognized and isn't limited to very few signs like many other papers. The number of signs used in this paper for classification is 28, which are used all around the globe. Two separate neural networks have been used for detection and recognition purpose; one classifies the sign and other the shape. Image augmentation has been used to create the training and validation dataset. 40,000 images have been used to train the first classifier with 28000 positive images (images that contain the traffic signs) and 12000 negative images(images that do not contain any traffic signs) and 3600 images were used to train the second classifier with 2400 positive images and 1200 negative images. The images are processed to find the region of interest, which is then fed to two CNN classifiers for classification.