CONTEXT AWARE DEEP LEARNING FRAMEWORK FOR WEATHER RESILIENT TRAFFIC SIGN DETECTION

Abstract

Road accidents are a leading cause of fatalities worldwide, and one major factor is drivers failing to notice or obey traffic signs due to distractions, poor visibility, or weather conditions. Advanced Driver Assistance Systems (ADAS) exist in luxury vehicles, but they are often expensive and not optimized for developing regions such as India. This work presents a Real-Time Traffic Sign Detection System using deep learning with the YOLOv5 architecture, integrated with image preprocessing and post-processing modules to enhance accuracy in diverse road conditions. The proposed system is trained on publicly available traffic sign datasets and optimized for real-time performance on resource constrained devices. Experimental results demonstrate that the system achieves reliable detection performance, making it a promising step toward cost-effective road safety solutions. Furthermore, the modular design of the system enables easy integration with edge devices such as Raspberry Pi or NVIDIA Jetson, making it suitable for deployment in smart vehicles and intelligent transportation systems. The results indicate the potential of this approach to reduce human error, improve driver awareness, and contribute to the development of safer road infrastructures.