Research Article

Deep learning-based traffic sign recognition system: A two-stage R-CNN approach with linear regression performance analysis

Abstract

An essential part of intelligent transportation systems (ITS) are systems for detecting and recognizing traffic signs autonomous vehicle technology. This research presents a comprehensive approach to develop Systems for detecting and classifying traffic signs in real time that use cutting-edge computer vision techniques and deep learning methods. The study addresses the fundamental challenge of accurately identifying small traffic signs within complex road environments, which is essential for improving driver assistance systems and improving road safety. Our proposed algorithm uses a two-stage approach based on the R-CNN framework, modelling traffic sign detection as a regional classification problem. The system uses convolutional neural networks (CNNs) trained on extensive datasets to achieve robust pattern recognition capabilities. Key features include real-time processing using vehicle-mounted cameras, text-to speech integration for voice alerts, and comprehensive classification of various traffic sign types based on their unique colours, shapes, and symbols. The research systematically addresses regional variations in traffic sign designs and implements effective pre-processing techniques to handle challenging environmental conditions. The test Results show that the technology can deliver precise, timely warnings to drivers, reduce distraction-related accidents, and improve overall traffic flow. This work contributes to the advancement of advanced driver assistance systems and autonomous driving technologies.

Authors affiliation:

Priyadarshini Baskarn: Information Technology, SRM Bharathidasan College of Engineering Technology, Pudukkottai 622515, India.
Nirmala Devi Karunanithi: Computer Science, SRM Bharathidasan College of Engineering Technology, Pudukkottai 622515, India.
Kavipriya Annadurai: Artificial Intelligence and Data Science, SRM Bharathidasan College of Engineering Technology, Pudukkottai 622515, India.
Devasena Govindan: Electrical Communication Engineering, SRM Bharathidasan College of Engineering Technology, Pudukkottai 622515, India.

How To Cite: P. Baskarn, N.D. Karunanithi, K. Annadurai and D. Govindan. Deep learning-based traffic sign recognition system: A two stage R-CNN approach with linear regression performance analysis. Journal of Computers and Applications, 2(2):18–25, 2026. https://doi.org/10.58613/jca222