Autonomous vehicles and advanced driver assistance systems (ADAS) depend heavily on traffic sign detection. Rain, fog, snow, as well as low light levels, which deteriorate visual quality and mask sign features, make it difficult to achieve reliable detection in a variety of weather circumstances. In order to overcome these difficulties in dynamic situations, this study proposes a deep learning-powered weather-resilient traffic sign detection framework. The suggested system incorporates a multi-phase strategy that combines convolutional neural networks (CNNs), adaptive feature extraction approaches, and data augmentation techniques. The algorithm is trained on both synthetic along with real-world datasets to identify traffic signs in a variety of weather conditions. A region-based attention technique that highlights traffic sign regions while reducing irrelevant background noise and a weather-invariant feature extraction module are important components. Extensive trials show that the framework achieves great performance across a variety of benchmarks, including unfavorable weather circumstances, and surpasses current state-of-the-art algorithms for detection accuracy and robustness. The findings demonstrate how the suggested framework could improve road safety by facilitating accurate traffic sign identification in intricate, real-world settings.