Advancements in Traffic Sign Detection and Recognition for Adverse Image and Motion Artifacts in Transportation Systems

Traffic sign detection and recognition (TSDR) contributes to the development of autonomous vehicle technology and driver assistance systems. The development of a TSDR that not only demonstrates high accuracy but also demonstrates robustness and reliability in a wide range of real-world scenarios is a critical prerequisite for the safe and widespread adoption of this technology. Nonetheless, the difficulty lies in the significant variability existing among traffic signs that must be identified, as well as the less-than-ideal quality of traffic images obtained in uncontrolled conditions, often obscured by bad image and motion artifacts. So, this research focuses on improving TSDR-Net, a traffic sign recognition system for autonomous driverless vehicles that work in harsh image conditions. The method uses multiple steps. It starts with hazy traffic sign images and uses deep learning techniques, particularly Deep Learning Convolutional Neural Network (DLCNN), to remove haze. The TSDR-Net begins by gathering hazy traffic sign images that have been affected by bad image; using deep learning for haze removal to improve visibility; using a DLCNN to detect traffic signs; and evaluating model performance through accuracy and loss calculations. By addressing the crucial task of traffic sign recognition in harsh image conditions, this research improves the safety and efficiency of autonomous driving systems.