Proposed for hyperspectral images classification. Each 3D and dense consideration network
Proposed for hyperspectral photos classification. Each 3D and dense attention network is proposed for hyperspectral pictures classification. Each 3D and 2D CNNs are combined in an end-to-end network. Specifically, the 3D multibranch and 2D CNNs are combined in an end-to-end network. Particularly, the 3D multibranch feature fusion function fusion module is made to extract multiscale attributes in the the spatial specis made to extract multiscale capabilities from spatial and and trum of of hyperspectral images. Following that, a 2D 2D dense attention module is spectrumthe the hyperspectral pictures. Following that, adense attention module is introduced. The The module consists of a densely connected block as well as a spatial-channel introduced. module consists of a densely connected block plus a spatial-channel interest focus block. The dense block is intended to alleviate gradient vanishing in deepand enblock. The dense block is intended to alleviate gradient vanishing in deep layers layers and enhance the reuse of features. interest module contains the spatial focus block and hance the reuse of features. The The consideration module consists of the spatial focus block and also the spectral interest block. The two blocks can adaptively pick the discriminative the spectral consideration block. The two blocks can adaptively pick the discriminative feafeatures from the space and also the spectrum of redundant hyperspectral images. Combining tures from the space and also the spectrum of redundant hyperspectral images. Combining the the densely connected block and attentionblock can drastically improve the classification densely connected block and focus block can significantly strengthen the classification performance and accelerate the convergence from the network. The elaborate hybrid module overall performance and accelerate the convergence of the network. The elaborate hybrid module raises the OA by 0.93.75 on 4 different datasets. Also, the proposed model raises the OA by 0.93.75 on four distinct datasets. Additionally, the proposed model ML-SA1 Data Sheet outperforms other comparison approaches in terms of OA by 1.638.11 on the PU dataset, outperforms other comparison strategies when it comes to OA by 1.638.11 on the PU dataset, 0.266.06 on the KSC dataset, 0.763.48 on the SA dataset, and 0.463.39 around the 0.266.06 on the KSC dataset, 0.763.48 around the SA dataset, and 0.463.39 around the Grass_dfc_2013 dataset. These experimental results demonstrate that the model proposed Grass_dfc_2013 dataset. These experimental benefits demonstrate that the model proposed can attain PSB-603 Biological Activity satisfactory classification functionality. can achieve satisfactory classification efficiency.Author Contributions: Y.Z. (Yiyan Zhang) and H.G. conceived the suggestions; Z.C., C.L. and Y.Z. (Yunfei Author Contributions: Y.Z. (Yiyan Zhang) and H.G. conceived the ideas; Z.C., C.L., and Y.Z. (YunZhang) gavegave suggestions for improvement; (Yiyan Zhang) and H.G. performed the experiment fei Zhang) ideas for improvement; Y.Z. Y.Z. (Yiyan Zhang) and H.G. carried out the experiand compiled the paper. H.Z. assisted and revisedrevised the All authorsauthors have read and for the ment and compiled the paper. H.Z. assisted plus the paper. paper. All have study and agreed agreed published version version of the manuscript. to the published in the manuscript. Funding: This operate is supported by National Organic Science Foundation of China (62071168), NatFunding: This operate is supported by National Natural Science Foundatio.
