This project is designed to enhance the capabilities of deep learning models in recognizing and classifying traffic signs. The dataset is the German Traffic Sign Benchmark (GTSRB), which consists of over 50,000 images, each annotated with one of 43 distinct traffic sign classes, representing a wide variety of sign types encountered on German roads. The project involves preprocessing the images, including normalization and augmentation techniques, to ensure robust model training. State-of-the-art convolutional neural networks (CNNs) are employed to extract spatial hierarchies and learn discriminative features essential for accurate classification. The model's performance is evaluated based on its ability to generalize across diverse lighting conditions, occlusions, and varying sign orientations, ultimately aiming to achieve high accuracy and reliability in real-world traffic sign recognition applications.