Graph Convolutional Networks for Hyperspectral Image Classification

By Danfeng Hong, Lianru Gao, Jing Yao, Bing Zhang, Antonio J. Plaza, Jocelyn Chanussot. 2020

1,248 citations recorded by the source corpus.

Convolutional neural networks (CNNs) have been attracting increasing attention in hyperspectral (HS) image classification due to their ability to capture spatial–spectral feature representations. Nevertheless, their ability in modeling relations between the samples remains limited. Beyond the limitations of grid sampling, graph convolutional networks (GCNs) have been recently proposed and successfully applied in irregular (or nongrid) data representation and analysis. In this article, we thoroughly investigate CNNs and GCNs (qualitatively and quantitatively) in terms of HS image classification. Due to the construction of the adjacency matrix on all the data, traditional GCNs usually suffer from a huge computational cost, particularly in large-scale remote sensing (RS) problems. To this end, we develop a new minibatch GCN (called miniGCN hereinafter), which allows to train large-scale GCNs in a minibatch fashion. More significantly, our miniGCN is capable of inferring out-of-sample data without retraining networks and improving classification performance. Furthermore, as CNNs and GCNs can extract different types of HS features, an intuitive solution to break the performance bottleneck of a single model is to fuse them. Since miniGCNs can perform batchwise network training (enabling the combination of CNNs and GCNs), we explore three fusion strategies: additive fusion, elementwise multiplicative fusion, and concatenation fusion to measure the obtained performance gain. Extensive experiments, conducted on three HS data sets, demonstrate the advantages of miniGCNs over GCNs and the superiority of the tested fusion strategies with regard to the single CNN or GCN models. The codes of this work will be available at https://github.com/danfenghong/IEEE_TGRS_GCN for the sake of reproducibility.

Subfields: Computer Vision and Pattern Recognition, Ecology, Media Technology, Graph Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Ecology

In Media Technology

Citations in the corpus

3D Instance Segmentation of MVS Buildings (2021, background): “into hyperspectral image classification in a minibatch fashion [38] and also propose a new transform-based network that”
Attention-Based Adaptive Spectral–Spatial Kernel ResNet for Hyperspectral Image Classification (2020, methodology): “[49] introduced graph convolution network (GCN) to model relations between samples for HSI classification.”
UIU-Net: U-Net in U-Net for Infrared Small Object Detection (2022, neutral): “learning-based methods [16], [17], especially deep learning methods, has rapidly made them the most widely used”
Self-Supervised Pretraining of Transformers for Satellite Image Time Series Classification (2020, background): “Various deep neural network architectures have been introduced to advance the state of the art for many remote sensing classification problems [16-18].”
MambaHSI: Spatial–Spectral Mamba for Hyperspectral Image Classification (2025, methodology): “DL-based be roughly divided into GCN-based [11], [13], CNN-based [15], [31], and Transformer-based [32], [33] methods.”
Deep learning for change detection in remote sensing: a review (2022, background): “…(Bazi et al. 2019), Unet (Jiao et al. 2020), Graph Convolutional Network (GCN) (Hong et al. 2020a; Gao et al. 2021), Spectral Former (Hong et al. 2021), and multimodal deep learning framework (Hong et al. 2020), which have been extensively used for remote sensing, are the potential models for DLCD.”
Extended Vision Transformer (ExViT) for Land Use and Land Cover Classification: A Multimodal Deep Learning Framework (2023, background): “Hong et al. [21] addressed the problem of large-scale graph computation burden in leveraging non-grid relation regularization and proposed a mini version of graph convolutional network for efficient HS image classification.”
SpectralFormer: Rethinking Hyperspectral Image Classification With Transformers (2021, background): “random forest (RF), 1-D CNN [14], 2-D CNN [17], RNN [18], miniGCN [21], transformers [22], and the proposed SpectralFormer.”
Multimodal Fusion Transformer for Remote Sensing Image Classification (2022, neutral): “Hence, CNN1D, CNN2D, and CNN3D produce relatively smooth classification maps where land-use and land-cover class boundaries are well separable.”
Hyperspectral Image Classification—Traditional to Deep Models: A Survey for Future Prospects (2021, methodology): “Translation: Encoder-Decoder Approaches [324] (GRU), Deep Convolutional Neural Networks for HSIC [218] (CNN1D), Deep Supervised Learning for Hyperspectral Data Classification through Convolutional Neural Networks [325] (CNN2D), 3-D Deep Learning Approach for Remote Sensing Image Classification [326]”
Spectral–Spatial Feature Tokenization Transformer for Hyperspectral Image Classification (2022, background): “generative adversarial networks (GANs) [51], [52], capsule networks [53], and graph convolutional networks (GCNs) [54] have also been introduced for HSI classification.”
Deep Learning for Unmanned Aerial Vehicle-Based Object Detection and Tracking: A survey (2021, background): “, DL) techniques [13]–[15], hierarchical feature representations with enough sample data can be learned with deep and complex networks.”