• Version
• Download 7
• File Size 431.42 KB
• File Count 1
• Create Date 10 July 2025
• Last Updated 10 July 2025

A Convolutional Approach to Traffic Sign Detection

B. KUMARI, G. BALA SOWMYA

Assistant Professor, 2 MCA Final Semester, Master of Computer Applications, Sanketika Vidya Parishad Engineering College, Vishakhapatnam, Andhra Pradesh, India.

Abstract:
A Convolutional Approach to Traffic Sign Detection presents a deep learning-based method using Convolutional Neural Networks (CNNs) for accurate traffic sign detection and classification. The system follows a two-stage pipeline: region proposal and sign classification, leveraging data augmentation and transfer learning for

improved performance. It demonstrates robustness to challenges like poor lighting and occlusion and supports real-time operation, making it suitable for ADAS and autonomous vehicles. Experimental results show high

accuracy across datasets, highlighting its potential in enhancing road safety and intelligent transportation systems.

Index Terms — Traffic Sign Detection, Convolutional Neural Networks (CNN), Deep Learning, Computer Vision, GTSRB, Image Classification, Object Detection, Real-Time Detection, Autonomous Driving, Data Augmentation, OpenCV, Keras, TensorFlow, Image Preprocessing, Road Safety