FUSED-Net: Detecting traffic signs with limited data
PLOS One · 2026
DOI: 10.1371/journal.pone.0359293 ↗Kurz gefasst
Fachartikel in der Zeitschrift PLOS One, 2026. Verfasst von Md Atiqur Rahman und 5 weiteren Autor:innen. Beteiligte Einrichtung: Department of Computer Science and Engineering, Islamic University of Technology, Board Bazar, Gazipur, Dhaka, Bangladesh. Hauptthemen laut MEDLINE: Neural Networks, Computer, Transportation, Pattern Recognition, Automated. Die Arbeit nennt 83 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Islamic University of Technology.
Abstract (Originalsprache)
Automatic Traffic Sign Recognition is paramount in modern transportation systems. However, curating large-scale datasets for diverse traffic sign detection remains impractical. In this context, we present FUSED-Net, a novel approach that enhances Few-Shot Object Detection (FSOD) for traffic signs using limited data. FUSED-Net integrates F aster RCNN with U nfrozen Parameters, Pseudo- S upport Sets, E mbedding Normalization, and D omain Adaptation to improve detection accuracy. Unlike conventional methods, FUSED-Net keeps all parameters unfrozen during training, enabling it to learn effectively from limited samples. A Pseudo-Support Set is generated through data augmentation, enhancing performance by compensating for the scarcity of target domain data. Embedding Normalization reduces intra-class variance, standardizing feature representations. Domain Adaptation, achieved by pre-training on a diverse traffic sign dataset, improves model generalization. Experimental results on the BDTSD dataset demonstrate that FUSED-Net achieves 2.4 × , 2.2 × , 1.5 × , and 1.3 × improvements in mAP under 1-shot, 3-shot, 5-shot, and 10-shot scenarios, respectively, compared to state-of-the-art FSOD models. Additionally, FUSED-Net achieves superior performance on the cross-domain FSOD benchmark across multiple settings. The source code and the URLs to download the datasets are available at https://github.com/180041123-Atiq/FUSED-Net .
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- Department of Computer Science and Engineering, Islamic University of Technology, Board Bazar, Gazipur, Dhaka, Bangladesh.
- TigerIT Bangladesh Ltd., Banani, Dhaka, Bangladesh.
- Department of Computer Science and Engineering, United International University, Badda, Dhaka, Bangladesh.
ORCID-Profile: Md Bakhtiar Hasan, Sabbir Ahmed, Md Hasanul Kabir
Förderung
- Islamic University of Technology
Zitieren
APA
Rahman, M. A., Asad, N., Haque, M. M., Hasan, M. B., Ahmed, S., & Kabir, M. H. (2026). FUSED-Net: Detecting traffic signs with limited data. PLOS One. https://doi.org/10.1371/journal.pone.0359293
Vancouver
Rahman MA, Asad N, Haque MM, Hasan MB, Ahmed S, Kabir MH. FUSED-Net: Detecting traffic signs with limited data. PLOS One. 2026. doi:10.1371/journal.pone.0359293
BibTeX
@article{Rahman2026,
author = {Rahman, Md Atiqur and Asad, Nahian Ibn and Haque, Md Mushfiqul and Hasan, Md Bakhtiar and Ahmed, Sabbir and Kabir, Md Hasanul},
title = {FUSED-Net: Detecting traffic signs with limited data},
journal = {PLOS One},
year = {2026},
doi = {10.1371/journal.pone.0359293}
}
EUVIMED