SpectralTrain: A Universal Framework for Hyperspectral Image Classification

2025/11/20 by Meihua Zhou, Lu Yu, Zhou, Meihua +12
Engineering · Environmental Science · Computer Science · #Remote-Sensing Image Classification #Remote Sensing in Agriculture #Machine Learning and Data Classification

paper · doi:10.48550/arxiv.2511.16084

Abstract

Hyperspectral image (HSI) classification typically involves large-scale data and computationally intensive training, which limits the practical deployment of deep learning models in real-world remote sensing tasks. This study introduces SpectralTrain, a universal, architecture-agnostic training framework that enhances learning efficiency by integrating curriculum learning (CL) with principal component analysis (PCA)-based spectral downsampling. By gradually introducing spectral complexity while preserving essential information, SpectralTrain enables efficient learning of spectral -- spatial patterns at significantly reduced computational costs. The framework is independent of specific architectures, optimizers, or loss functions and is compatible with both classical and state-of-the-art (SOTA) models. Extensive experiments on three benchmark datasets -- Indian Pines, Salinas-A, and the newly introduced CloudPatch-7 -- demonstrate strong generalization across spatial scales, spectral characteristics, and application domains. The results indicate consistent reductions in training time by 2-7x speedups with small-to-moderate accuracy deltas depending on backbone. Its application to cloud classification further reveals potential in climate-related remote sensing, emphasizing training strategy optimization as an effective complement to architectural design in HSI models. Code is available at https://github.com/mh-zhou/SpectralTrain.

Citations

• Dual Selective Fusion Transformer Network for Hyperspectral Image Classification
• Adam-mini: Use Fewer Learning Rates To Gain More
• Hyperspectral and multispectral image fusion with arbitrary resolution through self-supervised representations
• EfficientTrain++: Generalized Curriculum Learning for Efficient Visual Backbone Training
• 3D-Convolution Guided Spectral-Spatial Transformer for Hyperspectral Image Classification
• Cross-Entropy Loss Functions: Theoretical Analysis and Applications
• Weight Prediction Boosts the Convergence of AdamW
• DN-DETR: Accelerate DETR Training by Introducing Query DeNoising
• Masked Autoencoders Are Scalable Vision Learners
• Keypoint Transformer: Solving Joint Identification in Challenging Hands and Object Interactions for Accurate 3D Pose Estimation
• Rethinking Pre-training and Self-training
• Language Models are Few-Shot Learners
• Deep learning classifiers for hyperspectral imaging: A review
• Deep Residual Learning for Image Recognition
• BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

Related

• Advances in Hyperspectral Image Classification: Earth Monitoring with Statistical Learning Methods
• Curriculum Multi-Task Self-Supervision Improves Lightweight Architectures for Onboard Satellite Hyperspectral Image Segmentation
• HyperspectralMAE: The Hyperspectral Imagery Classification Model using Fourier-Encoded Dual-Branch Masked Autoencoder
• HSIMamba: Hyperpsectral Imaging Efficient Feature Learning with Bidirectional State Space for Classification
• A 3-stage Spectral-spatial Method for Hyperspectral Image Classification