Traffic Sign Improvement using AI

Traffic signs render crucial information about the upcoming road conditions, including details on turns, intersections, and other vital features. The current traffic detection systems struggle in complex traffic scenarios such as distorted signs, poor lighting, and adverse weather conditions. Incorrect interpretation of traffic signs by automotive vehicles causes injury, property damage, and fatalities. In this study image segmentation technique is implemented to segment the images. The difficulty arises in managing uncertainty in traffic sign images caused by fading clarity conditions. Furthermore, the distinction between background and foreground becomes complex amid adverse weather or other potential factors. Rough Set based Adaptive Fuzzy C-Means (RSAFCM) technique will be implemented, and the incorporation of rough set theory will handle the uncertainty and vagueness in the image due to complex conditions. Adaptive FCM will automatically tune its parameters to better suit the image characteristics, leading to improved clustering and accuracy of the segmentation results. This combined approach will facilitate precise cleaning and segmentation of the image. By employing the solution, we aim for accurate object and background segmentation. Further, this cleaned image is fed to the CNN for feature extraction and classification. Moreover, in this study 93% accuracy is achieved by employing the proposed methodology.