Copyright © 2025 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{181660,
        author = {Dr. Arpana Chaturvedi and Praveen Malik and Deepankar Jugran},
        title = {Traffic Sign Recognition Using Transfer Learning with Pre-trained CNNs and Bayesian Hyperparameter Optimization},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {12},
        number = {1},
        pages = {4847-4853},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=181660},
        abstract = {The visual appearance of traffic signs varies greatly in real-world settings, which is one of their defining characteristics. The perception of road signs is impacted, for instance, by variations in lighting, changing weather, and partial occlusions. In real-world applications, a huge variety of distinct sign classes must be accurately detected. Traffic signs are made to be simple to read by humans, who are particularly good at this task. However, identifying traffic signs still appears to be a difficult pattern recognition task for computer systems. Transfer learning's major objective is to transfer pertinent knowledge from the source domain while enhancing learning in the target domain. This research concentrates on transfer learning using deep convolutional neural network (CNN) and its designs, including VGG16, ResNet101 and EfficientNetB2, to handle such challenges. Then by applying Bayesian Optimization, we tuned the hyperparameters. In the study, these pre-trained CNN classifiers without their TOP layers are trained and tested on three different datasets as German Traffic Sign Recognition Benchmark (GTSRB), the Chinese Traffic Sign Recognition Benchmark (CHTSRB) and Indian Traffic Sign Dataset (ITSD). According to experimental findings, the suggested approach performed well in terms of metrics for evaluating the identification of traffic signs. The ResNet101 model performs better than all other implemented models on all three datasets, it gives the accuracy of 97.6537% on GTSRB, 96.7677% on CHTSD and 90.9547% on ITSD.},
        keywords = {Bayesian Optimization, Convolutional Neural Network, Deep Learning, Pre-trained Models,  Traffic Sign Recognition, Transfer Learning.},
        month = {June},
        }