Introducing an intelligent system for detecting traffic signs with deep learning to reduce road accidents

The intelligent system for detecting and recognition traffic signs detects traffic signs on its way and warns the driver by receiving images from the camera installed on the vehicle. These systems can also be used in self-driving and smart vehicles and usually have two main parts: sign recognition from other parts of the image and the other detecting and recognizing the type of traffic signs. In this article, traffic signs are identified using the power of convolutional networks. In fact, designing a system for detecting traffic signs is associated with many problems: Images taken may be noisy for a variety of reasons. Intensity and dimming of ambient light change the color of images. Or the signs may not be exactly the same as the standard defined, and most importantly most of the work has been done on foreign traffic signs. In this paper, a system is provided to detecting and recognition any of the traffic signs so that it can work on native and domestic datasets and consider the "power" and "speed" of detection, which are two important factors in smart vehicles. The performance results of this system show that the accuracy of traffic sign detection on training data and test data is 99% and one of the salient features of this system compared to similar tasks is the high speed of traffic sign detection in any type of weather and Light conditions and the presence of noise.