It is demonstrated that self-supervised spectral–spatial learning improves early disease detection and offers a scalable, data-efficient solution for hyperspectral crop monitoring.
Abstract
Wheat yellow rust is a major threat to global food security, causing yield losses of up to 70% if not detected early. Hyperspectral imaging enables pre-symptomatic detection, but extracting discriminative spectral–spatial features is challenging due to high dimensionality, redundancy, limited labelled data, and subtle disease signatures. This study proposes a unified transformer-based framework that integrates self-supervised masked autoencoder (MAE) pretraining with a SpectralFormer classifier. The MAE learns spectral–spatial representations by reconstructing masked spectral tokens, thereby providing effective initialisation for downstream supervised learning. The pretrained encoder is subsequently fine-tuned for disease classification, and the fully integrated model is evaluated on a real-world UAV-acquired hyperspectral dataset. Results show that the proposed MAE-SpectralFormer achieves 98.6% accuracy, 98.4% F1-score, and 0.957 Receiver Operating Characteristic - Area Under the Curve (ROC-AUC), outperforming the strongest Convolutional Neural Network (CNN) baseline (Inception-ResNet) by 5.4 percentage points in overall accuracy and 6.0 points in Rust-class F1-score, and exceeding the supervised SpectralFormer by 2.6 and 2.4 points, respectively. These findings demonstrate that self-supervised spectral–spatial learning improves early disease detection and offers a scalable, data-efficient solution for hyperspectral crop monitoring.
A lightweight deep learning model, the Dual Multi-Layer Perceptron Feature Fusion Network (DMLPFFN) as a hyperspectral-based approach to the four tomato disease stages, which includes healthy, asymptomatic, early and late infection is demonstrated.
Poonam Chaudhary, Sneha Kandacharam· Indian Journal of Agricultur...· 0 citations
Introduction Early-stage detection and classification of lettuce heat responses are essential for non-destructive phenotyping, yet conventional assessment mainly relies on visible symptoms and manual observation. Methods This study constructed a lettuce hyperspectral dataset comprising heat-sensitive and heat-tolerant varieties under control and high-temperature treatments, and proposed the Dynamic Selective Peak Transformer (DSPformer). DSPformer integrates edge-enhanced feature extraction, dynamic multi-scale spatial-spectral representation, Peak-k selective attention, and a confusion-aware dynamic focal loss to enhance discriminative features while reducing spectral redundancy, class imbalance, and inter-class confusion. Results Under the patch-level evaluation protocol, DSPformer achieved 96.22% accuracy, 95.55% recall, 96.35% precision, and 95.95% F1-score, outperforming the compared CNN- and Transformer-based models. Day-wise evaluation showed that DSPformer reached 82.61% accuracy on Day 1 and 96.55% on Day 3, before visible heat-stress symptoms appeared on Day 6. Under a plant-level partition protocol, DSPformer maintained robust performance with 93.76 +/- 0.49% accuracy. Additional evaluation on the Indian Pines benchmark further demonstrated the applicability of DSPformer to general hyperspectral image classification. Discussion These findings suggest that hyperspectral imaging can capture heat-stress-sensitive information beyond visual phenotypes, and that DSPformer provides a promising framework for early, non-destructive lettuce heat-response screening and hyperspectral phenotyping-assisted breeding.
Minglu Tian, Hui Yang, Mengen Yuan et al.· Frontiers in Plant Science· 0 citations
The multimodal fusion framework is introduced to overcome the limitations of unimodal approaches and integrates the most appropriate data sources, thus allowing the earliest and most accurate detection of plant pathogens.
M. T. Qasim, L. A. Hameed, Zainab I. Mohammed et al.· Current Applied Science and...· 0 citations
An enhanced hybrid deep-learning method by combining graph neural networks (GNNs) and multi-layer perceptrons (MLPs) for effective strawberry disease detection in real environments of fields offers an accurate and explainable solution that has a computationally efficient commitment for real-time monitoring of disease in smart agriculture settings, particularly on low-cost hardware assets.
V. Bhosale, Chin-Shiuh Shieh· International Journal of Inf...· 0 citations
Wheat gluten index is an important indicator of food quality inspection. Although hyperspectral technology offers non-destructive and rapid detection potential, its application faces challenges including high-dimensional data redundancy, multicollinearity among adjacent wavelengths, and model overfitting risks in small-sample scenarios. This study proposes a rapid prediction method for wheat gluten index integrating adaptive spectral preprocessing with machine learning. Based on 89 wheat flour samples, adaptive wavelet denoising (average SNR improvement: 11.92 dB), successive projections algorithm (SPA) feature selection, and variance inflation factor (VIF) collinearity diagnosis reduced 2001 spectral features to 5 core variables (587 nm, 436nm_square, 1241 nm, 1080nm_square, 940nm_square). Ten algorithms including PLSR, SVR, XGBoost, and CatBoost were systematically compared under 70% training/30% test split with 10-fold cross-validation. CatBoost achieved optimal performance (R2 = 0.8552, RMSE = 7.8602), surpassing PLSR (R2 = 0.7090) and XGBoost (R2 = 0.7923) by 20.6% and 8.0% respectively. Notably, the well-tuned single CatBoost model outperformed stacking ensemble methods (R2 = 0.7803), providing empirical evidence for model selection in small-sample spectral analysis. SHAP interpretability analysis identified critical spectral bands corresponding to protein and starch absorption characteristics, offering guidance for portable equipment optimization and mechanistic understanding of spectral-quality relationships.
Shiyou Zhu, Juan Bai, Yu-Long Chen et al.· Food Chemistry· 0 citations