EUVIMEDEUVIMED entdecken
Crossref

HACI-Net: a hybrid attention and channel interaction network for food image recognition

Zhiyong Xiao, Chaoliang Liu, Zhaohong Deng

Multimedia Systems · 2026-11 · 32 (10)

DOI: 10.1007/s00530-026-02719-7 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Multimedia Systems, online erschienen am 16.10.2026. Verfasst von Zhiyong Xiao und 2 weiteren Autor:innen. Die Arbeit nennt 52 Literaturangaben. Gefördert von: Natural Science Foundation of China, Natural Science Foundation of Jiangsu Province.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Dietary management plays a crucial role in applications such as intelligent food services, nutritional assessment, and health monitoring. As a key enabling technology, food image recognition provides foundational support for automated dietary recognition and nutritional analysis. However, this task remains highly challenging due to the vast diversity of food categories, subtle visual differences among dishes, complex shooting conditions, and strong background interference. To address these challenges, this paper proposes a Hybrid Attention and Channel Interaction Network (HACI-Net) based on the ConvNeXt architecture. Building upon the efficient feature extraction capability of ConvNeXt, the proposed model integrates a hybrid attention module and a channel interaction module. The hybrid attention module combines spatial attention, coordinate attention, and channel attention mechanisms, and employs residual connections to stabilize training, effectively suppressing background noise and enhancing focus on salient food regions. Meanwhile, the channel interaction module leverages pointwise grouped convolutions and channel shuffling to facilitate information exchange across channels, thereby improving the discriminative power of the learned features. Experimental results demonstrate that the proposed HACI-Net achieves recognition accuracies of 94.15% and 85.48% on the VireoFood-172 and ChineseFoodNet datasets, respectively, outperforming current state-of-the-art models and validating its superior performance in food image recognition tasks.

Förderung

  • Natural Science Foundation of China
  • Natural Science Foundation of Jiangsu Province

Zitieren

APA

Xiao, Z., Liu, C., & Deng, Z. (2026). HACI-Net: a hybrid attention and channel interaction network for food image recognition. Multimedia Systems, 32(10). https://doi.org/10.1007/s00530-026-02719-7

Vancouver

Xiao Z, Liu C, Deng Z. HACI-Net: a hybrid attention and channel interaction network for food image recognition. Multimedia Systems. 2026;32(10). doi:10.1007/s00530-026-02719-7

BibTeX
@article{Xiao2026,
  author = {Xiao, Zhiyong and Liu, Chaoliang and Deng, Zhaohong},
  title = {HACI-Net: a hybrid attention and channel interaction network for food image recognition},
  journal = {Multimedia Systems},
  year = {2026},
  volume = {32},
  number = {10},
  doi = {10.1007/s00530-026-02719-7}
}
Kostenlos starten →