Road Sign Detection in Extreme Weather Conditions

Jianhui Zhang, Kun Wei, Wenqing Wei, Murat Muzdybayev
Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing, PMLR 278:270-277, 2025.

Abstract

In road traffic sign detection, the low detection precision of road traffic signs in the detection screen is attributed to their small proportion and adverse weather conditions, such as fog, snow, and nighttime. To enhance the detection precision of road signs in extreme weather, this paper proposes an algorithm based on a lightweight improvement of YOLOv8n, referred to as FRPP-YOLOv8n (FFA-Net-RFAConv-PSA-P2-YOLOv8n). Firstly, the improved lightweight FFA-Net module is incorporated to dehaze the images. Secondly, RFAConv (Receptive-Field Attention convolutional operation) is introduced to enhance network performance. The PSA (Partial self-attention) mechanism is employed to improve detection precision, and finally, a small target detection layer is added to enhance the detection precision of small targets. The experimental results indicate that the improved algorithm achieves 53.4% mAP50-95 and 82.1% mAP50 on the CCTSDB2021 traffic sign dataset, which is an increase of 4.5% for both metrics compared to the original algorithm. Additionally, it maintains a high precision rate of 91.3%, representing a 4% improvement over the original algorithm.

Cite this Paper

BibTeX
@InProceedings{pmlr-v278-zhang25d, title = {Road Sign Detection in Extreme Weather Conditions}, author = {Zhang, Jianhui and Wei, Kun and Wei, Wenqing and Muzdybayev, Murat}, booktitle = {Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing}, pages = {270--277}, year = {2025}, editor = {Zeng, Nianyin and Pachori, Ram Bilas and Wang, Dongshu}, volume = {278}, series = {Proceedings of Machine Learning Research}, month = {25--27 Apr}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v278/main/assets/zhang25d/zhang25d.pdf}, url = {https://proceedings.mlr.press/v278/zhang25d.html}, abstract = {In road traffic sign detection, the low detection precision of road traffic signs in the detection screen is attributed to their small proportion and adverse weather conditions, such as fog, snow, and nighttime. To enhance the detection precision of road signs in extreme weather, this paper proposes an algorithm based on a lightweight improvement of YOLOv8n, referred to as FRPP-YOLOv8n (FFA-Net-RFAConv-PSA-P2-YOLOv8n). Firstly, the improved lightweight FFA-Net module is incorporated to dehaze the images. Secondly, RFAConv (Receptive-Field Attention convolutional operation) is introduced to enhance network performance. The PSA (Partial self-attention) mechanism is employed to improve detection precision, and finally, a small target detection layer is added to enhance the detection precision of small targets. The experimental results indicate that the improved algorithm achieves 53.4% mAP50-95 and 82.1% mAP50 on the CCTSDB2021 traffic sign dataset, which is an increase of 4.5% for both metrics compared to the original algorithm. Additionally, it maintains a high precision rate of 91.3%, representing a 4% improvement over the original algorithm.} }
Endnote
%0 Conference Paper %T Road Sign Detection in Extreme Weather Conditions %A Jianhui Zhang %A Kun Wei %A Wenqing Wei %A Murat Muzdybayev %B Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing %C Proceedings of Machine Learning Research %D 2025 %E Nianyin Zeng %E Ram Bilas Pachori %E Dongshu Wang %F pmlr-v278-zhang25d %I PMLR %P 270--277 %U https://proceedings.mlr.press/v278/zhang25d.html %V 278 %X In road traffic sign detection, the low detection precision of road traffic signs in the detection screen is attributed to their small proportion and adverse weather conditions, such as fog, snow, and nighttime. To enhance the detection precision of road signs in extreme weather, this paper proposes an algorithm based on a lightweight improvement of YOLOv8n, referred to as FRPP-YOLOv8n (FFA-Net-RFAConv-PSA-P2-YOLOv8n). Firstly, the improved lightweight FFA-Net module is incorporated to dehaze the images. Secondly, RFAConv (Receptive-Field Attention convolutional operation) is introduced to enhance network performance. The PSA (Partial self-attention) mechanism is employed to improve detection precision, and finally, a small target detection layer is added to enhance the detection precision of small targets. The experimental results indicate that the improved algorithm achieves 53.4% mAP50-95 and 82.1% mAP50 on the CCTSDB2021 traffic sign dataset, which is an increase of 4.5% for both metrics compared to the original algorithm. Additionally, it maintains a high precision rate of 91.3%, representing a 4% improvement over the original algorithm.
APA
Zhang, J., Wei, K., Wei, W. & Muzdybayev, M.. (2025). Road Sign Detection in Extreme Weather Conditions. Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing, in Proceedings of Machine Learning Research 278:270-277 Available from https://proceedings.mlr.press/v278/zhang25d.html.

Related Material