NL Journal of Agriculture and Biotechnology
(ISSN: 3048-9679)

---

AFD-Net: Wheat Foliar Disease Multi-Classification Using Deep Learning on Plant Pathology Datasets – A Comprehensive Review

Author(s) : Tsigehana Yewste Mamo*.
DOI : 10.71168/NAB.03.01.141

---

Abstract

Wheat foliar diseases such as leaf rust, stripe rust, tan spot, and powdery mildew continue to threaten global wheat production, necessitating rapid and accurate diagnostic tools to support timely disease management. Recent advances in deep learning have enabled high-performance image-based detection systems, with AFD-Net (Attention and Feature-Distilled Network) emerging as a state-of-the-art framework for fine-grained wheat disease classification. This review synthesizes current knowledge on AFD-Net architectures, training strategies, benchmark datasets, evaluation metrics, and real-world applicability. We highlight how the integration of attention mechanisms, multi-scale feature extraction, and feature distillation enables the model to capture subtle visual differences among closely related foliar diseases. Comparative analysis with existing convolutional neural networks and transformer-based models demonstrates the superior accuracy, robustness, and computational efficiency of AFD-Net across multiple plant pathology datasets. Additionally, we discuss practical deployment pathways including mobile-based diagnosis, drone-assisted crop monitoring, breeding program integration, and disease forecasting systems along with existing challenges related to dataset quality, environmental variability, model generalization, and explainability. Finally, future research directions are proposed to enhance model interpretability, dataset diversity, multimodal fusion, and transfer learning capabilities. Overall, AFD-Net represents a significant advancement in automated wheat foliar disease detection and holds strong potential for supporting sustainable and precise crop protection. Keywords: AFD-Net, Wheat foliar diseases, Deep learning, Attention mechanisms, Feature distillation, Image-based disease detection, Precision agriculture, multi-class classification.

This article licensed under the Creative Commons Attribution 4.0 International License CC-BY 4.0., which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.