Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Self-supervised Fusarium Head Blight Detection with Hyperspectral Image and Feature Mining

Yu-Fan Lin, Ching-Heng Cheng, Bo-Cheng Qiu, Cheng-Jun Kang, Chia-Ming Lee, Chih-Chung Hsu

TL;DR

Fusarium Head Blight detection is critical for cereal crop security but traditionally relies on labor-intensive manual scouting. The paper proposes a self-supervised hyperspectral imaging approach that uses endmember extraction guided by top-K spectral bands, with K-means pseudo-labels to mine discriminative features, followed by a simple classifier (LightGBM) on the selected bands. The method achieves robust discrimination between mild- and serious-FHB while avoiding deep networks and large labeled datasets, and identifies key spectral regions in the 700–750 nm and 800–875 nm ranges. This approach reduces dimensionality and computational demands, offering a practical solution for large-scale agricultural monitoring and deployment, demonstrated on the Beyond Visible Spectrum Challenge 2024 dataset. The work provides actionable insights into band-level spectral importance for FHB detection and a GPU-free, scalable pipeline for real-world use.

Abstract

Fusarium Head Blight (FHB) is a serious fungal disease affecting wheat (including durum), barley, oats, other small cereal grains, and corn. Effective monitoring and accurate detection of FHB are crucial to ensuring stable and reliable food security. Traditionally, trained agronomists and surveyors perform manual identification, a method that is labor-intensive, impractical, and challenging to scale. With the advancement of deep learning and Hyper-spectral Imaging (HSI) and Remote Sensing (RS) technologies, employing deep learning, particularly Convolutional Neural Networks (CNNs), has emerged as a promising solution. Notably, wheat infected with serious FHB may exhibit significant differences on the spectral compared to mild FHB one, which is particularly advantageous for hyperspectral image-based methods. In this study, we propose a self-unsupervised classification method based on HSI endmember extraction strategy and top-K bands selection, designed to analyze material signatures in HSIs to derive discriminative feature representations. This approach does not require expensive device or complicate algorithm design, making it more suitable for practical uses. Our method has been effectively validated in the Beyond Visible Spectrum: AI for Agriculture Challenge 2024. The source code is easy to reproduce and available at {https://github.com/VanLinLin/Automated-Crop-Disease-Diagnosis-from-Hyperspectral-Imagery-3rd}.

Self-supervised Fusarium Head Blight Detection with Hyperspectral Image and Feature Mining

TL;DR

Fusarium Head Blight detection is critical for cereal crop security but traditionally relies on labor-intensive manual scouting. The paper proposes a self-supervised hyperspectral imaging approach that uses endmember extraction guided by top-K spectral bands, with K-means pseudo-labels to mine discriminative features, followed by a simple classifier (LightGBM) on the selected bands. The method achieves robust discrimination between mild- and serious-FHB while avoiding deep networks and large labeled datasets, and identifies key spectral regions in the 700–750 nm and 800–875 nm ranges. This approach reduces dimensionality and computational demands, offering a practical solution for large-scale agricultural monitoring and deployment, demonstrated on the Beyond Visible Spectrum Challenge 2024 dataset. The work provides actionable insights into band-level spectral importance for FHB detection and a GPU-free, scalable pipeline for real-world use.

Abstract

Fusarium Head Blight (FHB) is a serious fungal disease affecting wheat (including durum), barley, oats, other small cereal grains, and corn. Effective monitoring and accurate detection of FHB are crucial to ensuring stable and reliable food security. Traditionally, trained agronomists and surveyors perform manual identification, a method that is labor-intensive, impractical, and challenging to scale. With the advancement of deep learning and Hyper-spectral Imaging (HSI) and Remote Sensing (RS) technologies, employing deep learning, particularly Convolutional Neural Networks (CNNs), has emerged as a promising solution. Notably, wheat infected with serious FHB may exhibit significant differences on the spectral compared to mild FHB one, which is particularly advantageous for hyperspectral image-based methods. In this study, we propose a self-unsupervised classification method based on HSI endmember extraction strategy and top-K bands selection, designed to analyze material signatures in HSIs to derive discriminative feature representations. This approach does not require expensive device or complicate algorithm design, making it more suitable for practical uses. Our method has been effectively validated in the Beyond Visible Spectrum: AI for Agriculture Challenge 2024. The source code is easy to reproduce and available at {https://github.com/VanLinLin/Automated-Crop-Disease-Diagnosis-from-Hyperspectral-Imagery-3rd}.
Paper Structure (6 sections, 1 equation, 5 figures)

This paper contains 6 sections, 1 equation, 5 figures.

Table of Contents

  1. Introduction
  2. Methodolodgy
  3. Top-k bands selection for Endmember Extraction
  4. Simple Classifier for Fusarium Head Blight Detection
  5. Experiment Results
  6. Conclusion

Figures (5)

  • Figure 1: The overall pipeline of the proposed method.
  • Figure 2: Elbow method (left) and Silhouette Analysis (right). Both methods show that when K is set to 2 as the best selection.
  • Figure 3: The overall normalized spectrum profiles of the FHB dataset.
  • Figure 4: The performance comparison of top-K bands selection strategy with different classifier and selected bands.
  • Figure 5: The indication of important bands on both serious-FHB and mild-FHB spectrum profiles.