In today's world, the number of vehicles is rising globally at a rapid pace, and road accidents are also increasing correspondingly. In Ethiopia, traffic accidents are becoming increasingly severe and alarming. In this research the high incidence of traffic accidents, often due to drivers overlooking or misinterpreting traffic signs, broken or distorted signs, obstruction by tree or other hindrance, damaged by human, and hand crafted classification fail. Contributing factors include driver fatigue, distractions, adverse weather conditions, and the manual, labor-intensive process of traffic sign detection. Support Vector Machines (SVMs) for traffic sign classification, emphasizing robustness to rotation and computational efficiency. By segmenting images based on color features, it detects blobs with colors commonly found in road signs, like red and blue, and then uses SVMs to classify these shapes (circular, triangular, rectangular, octagonal) by analyzing distances from edges.(Alcalá et al., 2004). To addressing could involve integrating neural networks to enhance content recognition accuracy and improve overall system effectiveness. The research proposed the use of Convolutional Neural Network (CNN) architectures, specifically VGG16 and InceptionV3, to classify traffic signs using images captured from Bahir Dar and its surrounding areas. The models are trained and tested using preprocessed (image resize, noise removal, contrast enhancement, segmentation) and augmented images, with InceptionV3 achieving a higher accuracy of 98.5% compared to VGG16's 97.3%. The study highlights the importance of automatic traffic sign classification in enhancing driver assistance systems, traffic management and autonomous vehicle particularly in challenging conditions where manual classification may fail. Key word: Deep Learning Techniques, Traffic Sign Classification.