Central Asian Research Index

Chemical Technology Control and Management, 2025, т. 2025, № 4, с. 81-89

EVALUATION OF DEEP LEARNING TECHNIQUES IN ROAD SIGN RECOGNITION

Latafat Abbas Gardashova, Haji Fakhraddin Hajiyev

Azerbaijan State Oil and Industry University. Address: Azadliq Avenue 34, AZ1010, Baku, Azerbaijan.

Аннотация · Abstract

Deep learning has transformed the computer vision field and greatly improved the performance and efficiency of road sign recognition systems. This research compares different deep learning methods, such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and hybrid models, in terms of their ability to effectively detect and classify road signs under various conditions. The study compares performance measures such as accuracy, processing speed, and robustness to environmental conditions like low lighting, occlusion, and adverse weather. The results show that CNN-based methods, especially those with transfer learning and ensemble techniques, have better performance in real-time scenarios. Problems like computational complexity and data quality issues are also discussed, as well as some possible solutions for deep learning model optimization for road sign detection. The paper emphasizes the necessity of the integration of such techniques into autonomous driving systems and intelligent transportation systems for the enhancement of road safety and traffic control. Furthermore, the research highlights the growing importance of scalable and adaptable architectures that can be efficiently deployed on embedded devices with limited computational resources. Ethical considerations such as algorithmic transparency, potential bias in datasets, and regulatory compliance are also touched upon. Future work may explore the fusion of multimodal data sources (e.g., LiDAR, GPS) to improve detection reliability and support context-aware decision-making in autonomous vehicles.

Сведения · Details