Light SDI-NAS: Lightweight Convolutional Neural Networks for Surface Defect Segmentation Based on Neural Architecture Search
Applied Sciences · 16.04.2026 · 16 (8) · 3875
DOI: 10.3390/app16083875 ↗Kurz gefasst
Fachartikel in der Zeitschrift Applied Sciences, online erschienen am 16.04.2026. Verfasst von Qingshi Chen und 3 weiteren Autor:innen. Die Arbeit nennt 38 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Hubei Engineering Research Center for Intelligent Detection and Identification of Complex Parts, Education Department of Hunan Province, the Project of State Key Laboratory of Precision Manufacturing for Extreme Service Performance.
Abstract (Originalsprache)
Convolutional neural networks (CNNs) have achieved remarkable performance in industrial image-based surface defect inspection in recent years. However, many state-of-the-art (SOTA) networks have become increasingly complex and computationally expensive, which limits their deployment in resource-constrained environments or high-throughput, real-time, practical industrial applications. To address this challenge, this paper proposes a novel approach, Light SDI-NAS, to automatically design lightweight CNN architectures for real-time industrial surface defect inspection through neural architecture search (NAS). First, a task-oriented search space for industrial image inspection is constructed by integrating prior knowledge of neural network architecture design with empirical observations. Second, a novel loss function is introduced to balance model accuracy and computational efficiency during the architecture search process. Finally, the lightweight networks generated by Light SDI-NAS demonstrate strong performance on three industrial image datasets. Experimental results show that the proposed models achieve comparable or superior accuracy to manually designed SOTA networks while significantly reducing the number of parameters and improving inference speed by 1.8 times, making them highly suitable for real-time industrial inspection applications.
Förderung
- Hubei Engineering Research Center for Intelligent Detection and Identification of Complex Parts
- Education Department of Hunan Province
- the Project of State Key Laboratory of Precision Manufacturing for Extreme Service Performance
Zitieren
APA
Chen, Q., Chen, B., Jia, X., & Li, K. (2026). Light SDI-NAS: Lightweight Convolutional Neural Networks for Surface Defect Segmentation Based on Neural Architecture Search. Applied Sciences, 16(8), 3875. https://doi.org/10.3390/app16083875
Vancouver
Chen Q, Chen B, Jia X, Li K. Light SDI-NAS: Lightweight Convolutional Neural Networks for Surface Defect Segmentation Based on Neural Architecture Search. Applied Sciences. 2026;16(8):3875. doi:10.3390/app16083875
BibTeX
@article{Chen2026,
author = {Chen, Qingshi and Chen, Biao and Jia, Xianshi and Li, Kai},
title = {Light SDI-NAS: Lightweight Convolutional Neural Networks for Surface Defect Segmentation Based on Neural Architecture Search},
journal = {Applied Sciences},
year = {2026},
volume = {16},
number = {8},
pages = {3875},
doi = {10.3390/app16083875}
}
EUVIMED