Published Paper Details:

Accurate traffic sign detection (TSD) is a critical component of autonomous driving and intelligent transportation systems, enabling vehicles to navigate safely and comply with traffic regulations. However, dynamic challenges such as small object detection, adverse weather conditions, and real-time performance constraints hinder the effectiveness of existing solutions. This research focuses on developing YOLOv7, a lightweight and efficient model optimized for real-time deployment on edge devices. The model incorporates innovative modules such as ELAN-REP for backbone optimization, CBAM for enhanced feature attention, AFPN for robust feature fusion, and SIOU loss for improved bounding box regression. To ensure real-world applicability, the Indian Traffic Sign Dataset (ITSD) was used, augmented with diverse environmental scenarios like rain, fog, and low light, to train and evaluate the model. Comparative analysis against state-of-the-art methods highlights the proposed model's superior trade-off between accuracy, speed, and computational efficiency. Real-world validation on edge devices such as NVIDIA Jetson Nano demonstrates the model's readiness for practical deployment. This research establishes a foundation for robust and efficient TSD in complex environments, paving the way for safer and more reliable autonomous driving technologies in diverse real-world scenarios.