Road Sign Detection Using R-CNN

The progression of automatous automobile navigation systems relies on the study of Traffic Sign Recognition (TSR). Because of the need to constantly analyze the situation to make decisions, all autonomous driving systems lack the capacity to run complex image processing algorithms as well as large-scale computing responsibilities. However, they need the efficient calculations of input data in order to generate outputs in the shortest time possible. This is due the fact that there are a lot of traffic signs, advertisements, parked cars, individuals and many other vehicles and objects both in motion and stationery in urban environments. While the strategies discussed in literature for the implementation of ITS have been proposed in varied forms, few have been modelled and analyzed at most at moderate speeds, or with regards to rural or urban roads. Hence, the aim of this paper is to give a background of the main problem and the available techniques to solve those issues, and present a detailed approach to tackle real-time recognition problems in urban traffic sign detection.