Hybrid Dense Network With Attention Mechanism for Hyperspectral Image Classification

Khan, Adil; Mazzara, Manuel; Distefano, Salvatore; Roy, Swalpa Kumar; Wu, X.

doi:10.1109/jstars.2022.3171586

Cited by 20 publications

(5 citation statements)

References 56 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The activation maps for the spatial-spectral position (x, y, z) at the i-th feature map and j-th layer can be denoted as v (x,y,z) i,j in which d i−1 represents the total number of feature maps at the (i − 1)-th layer, w i,j and b i,j denote the depth of the kernel and bias, respectively. Additionally, 2γ + 1, 2δ + 1, and 2ν + 1 correspond to the height, width, and depth of the kernel [19], [63]. The Swin Transformer (ST) excels in constructing multiscale feature maps by iteratively fusing neighboring patches using the window partition mechanism.…”

Section: Proposed Methodologymentioning

confidence: 99%

Groundwater Contamination Issues in Chiniot Area, Punjab, Pakistan

Ahmad¹,

Sultana²,

Yoshida³

et al. 2020

IJESD

View full text Add to dashboard Cite

This study was planned to assess the groundwater quality of the District Chiniot, Pakistan. The majority of the people in the area use the groundwater from private wells. Water quality is very poor in the area especially close to the Pre Cambrian exposed Hills. TDS (total dissolved solids), hardness, Chloride, Sulfate and Calcium were very high in groundwater. Trace element chemistry shows that As, Mn and S are above the recommended limit. As and Mn contamination is due to water-rock interaction. Whereas S is encountered in shallow wells, which may be due to improper solid waste management and septic tanks. Microbiological analysis of municipal supply water shows that 25% of water samples have faecal contamination. High TDS groundwater is the cause of economic loss due to damage of sanitary fittings.

show abstract

Section: Proposed Methodologymentioning

confidence: 99%

Groundwater Contamination Issues in Chiniot Area, Punjab, Pakistan

Ahmad¹,

Sultana²,

Yoshida³

et al. 2020

IJESD

View full text Add to dashboard Cite

show abstract

“…A fast and compact 3dimensional CNN model has been proposed in [52] which significantly reduces the computational cost and improves the experimental results for several Hyperspectral datasets. In this hierarchy, the works [37], [38], [53], [54] proposed Hybrid 3-dimensional followed by 2-dimensional CNN layers for a better spatial-spectral feature hierarchy for end classification.…”

Section: Literature Reviewmentioning

confidence: 99%

A Disjoint Samples-Based 3D-CNN With Active Transfer Learning for Hyperspectral Image Classification

Ghous

Hong

Khan

et al. 2022

IEEE Trans. Geosci. Remote Sensing

Self Cite

View full text Add to dashboard Cite

Convolutional Neural Networks (CNNs) have been extensively studied for Hyperspectral Image Classification (HSIC). However, CNNs are critically attributed to a large number of labeled training samples, which outlays high costs in terms of time and resources. Moreover, CNNs are trained on some samples and have been tested on the entire HSI. Perhaps, the entire HSI is taken into account at test time to appropriately generate the ground truth maps. In order to obtain a higher accuracy while considering the limited availability of training samples and disjoint validation and test samples, this work proposes a fast and compact 3D CNN-based Active Learning (AL) for HSIC that integrates both deep transfer learning and AL into a unified framework. In the proposed methodology, a 3D CNN model is trained with very few training samples (i.e., 5%, only) and in the next phase, the most informative and heterogeneous samples are queried from the validation set (candidate set) based on the fuzziness, mutual information and breaking ties of the trained model. The 3D CNN model is later fine-tuned (rather retraining from scratch) with the new training samples (i.e., 200 samples are selected in each iteration) to reduce the computational cost. The proposed method has been compared with the state-of-the-art traditional and deep models proposed for HSIC. Experimental results proved the superiority of our proposed method on several benchmark HSI datasets with significantly fewer labeled samples.Matlab demo can be accessed on GitHub: github.com/mahmad00

show abstract

“…Deep learning methods integrate feature extraction and classification into an end-to-end workflow and extract features automatically [11], thus dominating HSI classification. As a popular deep learning method, convolutional neural networks (CNNs) have been introduced into the HSI classification and demonstrated advanced performance [12][13][14]. However, the numerous spectral bands and complex crossband structure in HSI induce high-performance CNNs to become deeper and wider [15].…”

Section: Introductionmentioning

confidence: 99%

LiteCCLKNet: A lightweight criss‐cross large kernel convolutional neural network for hyperspectral image classification

Zhong

Gong²,

Zhang

et al. 2023

IET Computer Vision

View full text Add to dashboard Cite

High‐performance convolutional neural networks (CNNs) stack many convolutional layers to obtain powerful feature extraction capability, which leads to huge storing and computational costs. The authors focus on lightweight models for hyperspectral image (HSI) classification, so a novel lightweight criss‐cross large kernel convolutional neural network (LiteCCLKNet) is proposed. Specifically, a lightweight module containing two 1D convolutions with self‐attention mechanisms in orthogonal directions is presented. By setting large kernels within the 1D convolutional layers, the proposed module can efficiently aggregate long‐range contextual features. In addition, the authors effectively obtain a global receptive field by stacking only two of the proposed modules. Compared with traditional lightweight CNNs, LiteCCLKNet reduces the number of parameters for easy deployment to resource‐limited platforms. Experimental results on three HSI datasets demonstrate that the proposed LiteCCLKNet outperforms the previous lightweight CNNs and has higher storage efficiency.

show abstract

Hybrid Dense Network With Attention Mechanism for Hyperspectral Image Classification

Cited by 20 publications

References 56 publications

Groundwater Contamination Issues in Chiniot Area, Punjab, Pakistan

Groundwater Contamination Issues in Chiniot Area, Punjab, Pakistan

A Disjoint Samples-Based 3D-CNN With Active Transfer Learning for Hyperspectral Image Classification

LiteCCLKNet: A lightweight criss‐cross large kernel convolutional neural network for hyperspectral image classification

Contact Info

Product

Resources

About