EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Hierarchical Discriminative Feature Refinement for Robust and Lightweight Clothing Detection

Huawen Li, Kun Yang iD, Yingchao Wang iD, Le Chang, Feng Xue iD, Hui Li

Sensors · 30.10.2026 · 26 (21) · 6876

DOI: 10.3390/s26216876 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 30.10.2026. Verfasst von Huawen Li und 5 weiteren Autor:innen. Beteiligte Einrichtung: School of Fashion Design and Engineering, Zhejiang Sci-Tech University, Hangzhou 311199, China. Die Arbeit nennt 83 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Japan Science and Technology Agency, the Youth Fund of the Southern Xinjiang Joint Fund of the Natural Science Foundation of Xinjiang Uygur Autonomous Region, the Japan Science and Technology Agency (JST) through the Support for Pioneering Research Initiated by the Next Generation (SPRING) Program.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurate and efficient clothing detection remains challenging because visually similar garment categories can share appearance cues, while background interference, scale variation, and spatial overlap weaken category-discriminative features. To address these issues, this study proposes a lightweight clothing detector based on YOLO26n with hierarchical discriminative feature refinement. The framework progressively enhances garment representations at the channel, semantic, and multi-scale spatial levels. Specifically, the C3k2-based Channel Feature Enhancement Module (C3k2_CFEM) strengthens cross-channel interaction and adaptively balances local details with contextual information. The C2-based Gated Semantic Attention module (C2GSA) selectively retains task-relevant high-level responses while suppressing redundant contextual interference. The Multi-Scale Dual-Stream Attention Module (MSDSAM) combines complementary foreground–background attention with efficient multi-scale feature interaction. A multi-source clothing dataset containing 11,970 images and 21,504 annotated garment instances across seven categories was used for evaluation. The proposed model achieves 91.6% Precision, 86.2% Recall, 92.6% mAP50, and 74.7% mAP50:95, improving the YOLO26n baseline by 6.6 and 5.8 percentage points in mAP50 and mAP50:95, respectively. These results are obtained with 2.39 M parameters, 5.4 billion floating-point operations (GFLOPs), and an inference speed of 166 frames per second (FPS) on an NVIDIA RTX 4090 at 640×640 input resolution and batch size 1. Additional evaluations report the model’s behavior across clothing categories, target scales, overlap levels, image perturbations, hardware platforms, and external clothing datasets. Under the reported protocols, the framework improves detection accuracy with limited increases in parameters and computational cost.

Schlagwörter

Stichwörter der Autor:innen

YOLO26nclothing detectiongated attentionhierarchical discriminative feature refinementlightweight object detectionmulti-scale feature refinementrobust visual perception

Autor:innen und Institutionen

  • School of Fashion Design and Engineering, Zhejiang Sci-Tech University, Hangzhou 311199, China.
  • Graduate School of Information, Production and Systems, Waseda University, Kitakyushu 808-0135, Japan.
  • School of Intelligent Manufacturing and Electrical Engineering, Xinjiang College of Science & Technology, Korla 841000, China.
  • School of Fashion Art and Engineering, Beijing Institute of Fashion Technology, Beijing 100029, China.
  • Graduate School of Frontier Sciences, The University of Tokyo, Kashiwa 277-8561, Japan.

ORCID-Profile: Kun Yang, Yingchao Wang, Feng Xue

Förderung

  • Japan Science and Technology Agency
  • the Youth Fund of the Southern Xinjiang Joint Fund of the Natural Science Foundation of Xinjiang Uygur Autonomous Region
  • the Japan Science and Technology Agency (JST) through the Support for Pioneering Research Initiated by the Next Generation (SPRING) Program
  • Zhejiang University of Science and Technology
  • Youth Fund of the Southern Xinjiang Joint Fund of the Natural Science Foundation of Xinjiang Uygur Autonomous Region

Zitieren

APA

Li, H., Yang, K., Wang, Y., Chang, L., Xue, F., & Li, H. (2026). Hierarchical Discriminative Feature Refinement for Robust and Lightweight Clothing Detection. Sensors, 26(21), 6876. https://doi.org/10.3390/s26216876

Vancouver

Li H, Yang K, Wang Y, Chang L, Xue F, Li H. Hierarchical Discriminative Feature Refinement for Robust and Lightweight Clothing Detection. Sensors. 2026;26(21):6876. doi:10.3390/s26216876

BibTeX
@article{Li2026,
  author = {Li, Huawen and Yang, Kun and Wang, Yingchao and Chang, Le and Xue, Feng and Li, Hui},
  title = {Hierarchical Discriminative Feature Refinement for Robust and Lightweight Clothing Detection},
  journal = {Sensors},
  year = {2026},
  volume = {26},
  number = {21},
  pages = {6876},
  doi = {10.3390/s26216876}
}
Kostenlos starten →