Improving Traffic Signs Recognition Based Region Proposal and Deep Neural Networks

Nowadays, traffic sign recognition has played an important task in autonomous vehicle, intelligent transportation systems. However, it is still a challenging task due to the problems of a variety of color, shape, environmental conditions. In this paper, we propose a new approach for improving accuracy of traffic sign recognition. The contribution of this work is three-fold: First, region proposal based on segmentation technique is applied to cluster traffic signs into several sub regions depending upon the supplemental signs and the main sign color. Second, image augmentation of training dataset generates a larger data for deep neural network learning. This proposed task is aimed to address the small data problem. It is utilized for enhancing capabilities of deep learning. Finally, we design appropriately a deep neural network to image dataset, which combines the original images and proposal images. The proposed approach was evaluated on a benchmark dataset. Experimental evaluation on public benchmark dataset shows that the proposed approach enhances performance to 99.99% accuracy. Comparison results illustrated that our proposed method reaches higher performance than almost state-of-the-art methods.