Abstract:Supervised approaches classify input data using a set of representative samples for each class, known as training samples. The collection of such samples is expensive and time demanding. Hence, unsupervised feature learning, which has a quick access to arbitrary amounts of unlabeled data, is conceptually of high interest. In this paper, we propose a novel network architecture, fully Conv-Deconv network, for unsupervised spectral-spatial feature learning of hyperspectral images, which is able to be trained in an end-to-end manner. Specifically, our network is based on the so-called encoder-decoder paradigm, i.e., the input 3-D hyperspectral patch is first transformed into a typically lower dimensional space via a convolutional subnetwork (encoder), and then expanded to reproduce the initial data by a deconvolutional subnetwork (decoder). However, during the experiment, we found that such a network is not easy to be optimized. To address this problem, we refine the proposed network architecture by incorporating: 1) residual learning and 2) a new unpooling operation that can use memorized max-pooling indexes. Moreover, to understand the “black box,” we make an in-depth study of the learned feature maps in the experimental analysis. A very interesting discovery is that some specific “neurons” in the first residual block of the proposed network own good description power for semantic visual patterns in the object level, which provide an opportunity to achieve “free” object detection. This paper, for the first time in the remote sensing community, proposes an end-to-end fully Conv-Deconv network for unsupervised spectral-spatial feature learning. Moreover, this paper also introduces an in-depth investigation of learned features. Experimental results on two widely used hyperspectral data, Indian Pines and Pavia University, demonstrate competitive performance obtained by the proposed methodology compared with other studied approaches.

HyperKon: A Self-Supervised Contrastive Network for Hyperspectral Image Analysis

HyperKon: A Self-Supervised Contrastive Network for Hyperspectral Image Analysis

A Hyperspectral Image Classification Method Based on the Nonlocal Attention Mechanism of a Multiscale Convolutional Neural Network.

ContrastNet: Unsupervised Feature Learning by Autoencoder and Prototypical Contrastive Learning for Hyperspectral Imagery Classification.

Hierarchical Superpixel Relation Graph Combined with Convolutional Sparse Coding for Self-Supervised Hyperspectral Image Denoising

HyperKAN: Kolmogorov-Arnold Networks make Hyperspectral Image Classificators Smarter

Hyperspectral Image Classification With Contrastive Graph Convolutional Network

A hybrid approach consisting of 3D depthwise separable convolution and depthwise squeeze-and-excitation network for hyperspectral image classification

Active Deep Densely Connected Convolutional Network for Hyperspectral Image Classification

HyperRefiner: a Refined Hyperspectral Pansharpening Network Based on the Autoencoder and Self-Attention

HyperKAN: Kolmogorov–Arnold Networks Make Hyperspectral Image Classifiers Smarter

HDECGCN: A Heterogeneous Dual Enhanced Network Based on Hybrid CNNs Joint Multiscale Dynamic GCNs for Hyperspectral Image Classification

Self-Supervised Contrastive Learning Residual Network for Hyperspectral Image Classification under Limited Labeled Samples

HyperNet: Self-Supervised Hyperspectral Spatial-Spectral Feature Understanding Network for Hyperspectral Change Detection

Residual HSRCNN: Residual Hyper-Spectral Reconstruction CNN from an RGB Image

Unsupervised Spectral–Spatial Feature Learning via Deep Residual Conv–Deconv Network for Hyperspectral Image Classification

KNLConv: Kernel-space Non-local Convolution for Hyperspectral Image Super-resolution

Attention Graph Convolutional Network for Disjoint Hyperspectral Image Classification

Contrastive Graph Convolution Network With Skip Connections for Few-Shot Hyperspectral Image Classification

Hyperspectral Image Classification Based on a 3D Octave Convolution and 3D Multiscale Spatial Attention Network

HSCNN+: Advanced CNN-Based Hyperspectral Recovery from RGB Images