EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

PlantAIM: A new baseline model integrating global attention and local features for enhanced plant disease identification

Abel Yu Hao Chai, Sue Han Lee, Fei Siang Tay, Hervé Goëau, Pierre Bonnet, Alexis Joly

Smart Agricultural Technology · 2025-03 · 10 · 100813

DOI: 10.1016/j.atech.2025.100813 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Smart Agricultural Technology, 2025-03. Verfasst von Abel Yu Hao Chai und 5 weiteren Autor:innen. Die Arbeit wurde bisher 12-mal zitiert (Crossref) und nennt 69 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: Swinburne University of Technology - Sarawak Campus, Swinburne University of Technology, Agence nationale de la recherche.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Plant diseases significantly affect the quality and yield of agricultural production. Conventionally, detection has relied on plant pathologists, but recent advances in deep learning, particularly the Vision Transformer (ViT) and Convolutional Neural Network (CNN), have made it feasible for automated plant disease identification. Despite their prominence, there are still significant gaps in our understanding of how these models differ in feature extraction and representation, particularly in complex multi-crop disease identification tasks. This challenge arises from the simultaneous need to learn crop-specific and disease-specific features for accurate identification of crop species and its associated diseases. To address this, we introduce Plant Disease Global-Local Features Fusion Attention Model (PlantAIM), a new hybrid framework that fuses global attention mechanisms of ViT with local feature extraction capabilities of CNN. PlantAIM aims to improve the model's ability to simultaneously learn and focus on crop-specific and disease-specific features. We conduct extensive evaluations to assess the robustness and generalizability of PlantAIM compared to state-of-the-art (SOTA) models, including scenarios with limited training samples and real-world environmental data. Our results show that PlantAIM achieves superior performance. This research not only deepens our understanding of feature learning for ViT and CNN models, but also sets a new benchmark in the dynamic field of plant disease identification. The code is available at github: PlantAIM • Our proposed PlantAIM sets a new benchmark in multi-crop disease identification. • PlantAIM outperforms other SOTA models in multi-crop disease identification. • Self-attention and convolutional layers learn different features for plant disease. • PlantAIM's GLFA layer effectively fuses global and local features from ViT and CNN. • PlantAIM's analysis shows it captures crop and disease features concurrently.

Förderung

  • Swinburne University of Technology - Sarawak Campus
  • Swinburne University of Technology
  • Agence nationale de la recherche
  • Malaysia Ministry of Higher Education

Zitieren

APA

Chai, A. Y. H., Lee, S. H., Tay, F. S., Goëau, H., Bonnet, P., & Joly, A. (2025). PlantAIM: A new baseline model integrating global attention and local features for enhanced plant disease identification. Smart Agricultural Technology, 10, 100813. https://doi.org/10.1016/j.atech.2025.100813

Vancouver

Chai AYH, Lee SH, Tay FS, Goëau H, Bonnet P, Joly A. PlantAIM: A new baseline model integrating global attention and local features for enhanced plant disease identification. Smart Agricultural Technology. 2025;10:100813. doi:10.1016/j.atech.2025.100813

BibTeX
@article{Chai2025,
  author = {Chai, Abel Yu Hao and Lee, Sue Han and Tay, Fei Siang and Goëau, Hervé and Bonnet, Pierre and Joly, Alexis},
  title = {PlantAIM: A new baseline model integrating global attention and local features for enhanced plant disease identification},
  journal = {Smart Agricultural Technology},
  year = {2025},
  volume = {10},
  pages = {100813},
  doi = {10.1016/j.atech.2025.100813}
}
Kostenlos starten →