Traffic sign recognition (TSR) is a fundamental element of intelligent transport systems that plays a decisive role in improving road safety and optimizing traffic flow. The application of the network of progressive neurons (CNNs) to TSR has proved to be effective and potent as a means of improving smart transportation systems and road safety. This paper presents an intelligent system that uses CNN and YOLO (You Only Look Once) to effectively identify traffic signs, especially under challenging adverse weather conditions, characterized by low lighting and degraded image quality. Initially, the raw traffic scenes were acquired, followed by color-based segmentation to isolate the sign characteristics. Afterwards, CNNs were trained and evaluated by TSR using the German Traffic Sign Detection Benchmark (GTSDB) and GTSRB datasets. In addition, the deployment of CNNs is improved by the presence of extensive annotation datasets, which improves the network's ability to extrapolate to unfamiliar data.