Deep Learning-Based Traffic Banner Recognition for Enhanced Road Safety in Autonomous Vehicles

Abstract

Accurate recognition of traffic banners is crucial for the safety and reliability of modern intelligent transportation systems, including autonomous vehicles and advanced driver-assistance systems (ADAS). Traditional recognition methods often struggle under adverse conditions such as poor lighting, occlusions, and complex environmental variations, limiting their effectiveness in real-world applications. In this paper, we propose a robust, real-time traffic banner recognition model based on Convolutional Neural Networks (CNNs). Our model is optimized for high accuracy and computational efficiency on resource-constrained devices typical of autonomous vehicles. We introduce a comprehensive data preprocessing and augmentation pipeline that enhances the model’s robustness against challenging environmental conditions. Extensive experimental evaluations demonstrate that our model outperforms existing techniques in both accuracy and speed, highlighting its potential for seamless integration into intelligent transportation systems to enhance road safety.