Lightweight Improvement and Structural Applicability Analysis of YOLOv8n for Industrial Surface Defect Detection
Advances in Engineering Technology Research · 15.07.2026 · 17 (1) · 676
DOI: 10.56028/aetr.17.1.676.2026 ↗Kurz gefasst
Fachartikel in der Zeitschrift Advances in Engineering Technology Research, online erschienen am 15.07.2026. Verfasst von Huanhua Huang und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Industrial surface defect detection is constrained by large defect-scale variation, weak texture responses, blurred boundaries, and strict lightweight deployment requirements. To investigate the applicability of lightweight structural enhancement strategies, this study adopted YOLOv8n as the baseline detector and introduced a Lightweight Multi-scale Enhancement module, termed LME, at the feature-fusion stage. Extended variants integrating SRM, CSR, and P2 detection branches were further constructed under a unified experimental protocol. Comparative experiments, ablation studies, and visualization evaluation were conducted on the GC10-DET dataset. The results showed that YOLOv8n + LME achieved the best Precision, Recall, and mAP@0.5, reaching 66.53%, 65.49%, and 65.11%, respectively, indicating that LME improved coarse defect detection performance to a certain extent. However, its mAP@0.5:0.95 was lower than that of the baseline YOLOv8n, suggesting that the improvement was not fully transferred to stricter bounding-box localization accuracy. Further stacking CSR and P2 caused performance degradation. These results indicate that lightweight industrial defect detection should not rely solely on structural complexity, but should emphasize the matching relationship among feature-fusion design, defect characteristics, sample distribution, and localization requirements.
Zitieren
APA
Huang, H., & Wei, C. (2026). Lightweight Improvement and Structural Applicability Analysis of YOLOv8n for Industrial Surface Defect Detection. Advances in Engineering Technology Research, 17(1), 676. https://doi.org/10.56028/aetr.17.1.676.2026
Vancouver
Huang H, Wei C. Lightweight Improvement and Structural Applicability Analysis of YOLOv8n for Industrial Surface Defect Detection. Advances in Engineering Technology Research. 2026;17(1):676. doi:10.56028/aetr.17.1.676.2026
BibTeX
@article{Huang2026,
author = {Huang, Huanhua and Wei, Chunmiao},
title = {Lightweight Improvement and Structural Applicability Analysis of YOLOv8n for Industrial Surface Defect Detection},
journal = {Advances in Engineering Technology Research},
year = {2026},
volume = {17},
number = {1},
pages = {676},
doi = {10.56028/aetr.17.1.676.2026}
}
EUVIMED