Hyperspectral image classification using 3D 2D CNN
Abstract Recent works have shown that deep‐learning‐derived methods based on convolutional neural network can achieve high performance in terms of accuracy when applied to computer vision task such as object detection, segmentation and classification particularly on hyperspectral image. However, the...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2021-04-01
|
| Series: | IET Image Processing |
| Subjects: | |
| Online Access: | https://doi.org/10.1049/ipr2.12087 |
| _version_ | 1798035602904449024 |
|---|---|
| author | Alou Diakite Gui Jiangsheng Fu Xiaping |
| author_facet | Alou Diakite Gui Jiangsheng Fu Xiaping |
| author_sort | Alou Diakite |
| collection | DOAJ |
| description | Abstract Recent works have shown that deep‐learning‐derived methods based on convolutional neural network can achieve high performance in terms of accuracy when applied to computer vision task such as object detection, segmentation and classification particularly on hyperspectral image. However, the existing methods have long training times. To reduce the training time and increase the accuracy, this paper proposed a new 3D2D convolutional neural network combined for hyperspectral image classification. For this purpose, a 3D fast learning block (depthwise separable convolution block and a fast convolution block) followed by a 2D convolutional neural network was introduced to extract spectralspatial features. Four datasets were used for the experiment purpose and the results showed that the proposed method achieved excellent result on both small and large training data compared with existing methods. The proposed method increased the overall accuracies by 2% on UP and KSC datasets while significantly reducing the training time on IP and KSC datasets, respectively. The proposed method increased all accuracies for at least 6% on IP, KSC and UP datasets when compared to some state‐of‐the‐art methods. Also, it reduced considerably the training and testing time on IP and KSC datasets when fast convolution block alone is involved. |
| first_indexed | 2024-04-11T21:00:28Z |
| format | Article |
| id | doaj.art-bb66a7b3b6e74d1bb71a10feaae76a76 |
| institution | Directory Open Access Journal |
| issn | 1751-9659 1751-9667 |
| language | English |
| last_indexed | 2024-04-11T21:00:28Z |
| publishDate | 2021-04-01 |
| publisher | Wiley |
| record_format | Article |
| series | IET Image Processing |
| spelling | doaj.art-bb66a7b3b6e74d1bb71a10feaae76a762022-12-22T04:03:32ZengWileyIET Image Processing1751-96591751-96672021-04-011551083109210.1049/ipr2.12087Hyperspectral image classification using 3D 2D CNNAlou Diakite0Gui Jiangsheng1Fu Xiaping2School of Information Science Zhejiang Sci‐Tech University Hangzhou ChinaSchool of Information Science Zhejiang Sci‐Tech University Hangzhou ChinaFaculty of Mechanical Engineering and Automation Zhejiang Sci‐Tech University Hangzhou ChinaAbstract Recent works have shown that deep‐learning‐derived methods based on convolutional neural network can achieve high performance in terms of accuracy when applied to computer vision task such as object detection, segmentation and classification particularly on hyperspectral image. However, the existing methods have long training times. To reduce the training time and increase the accuracy, this paper proposed a new 3D2D convolutional neural network combined for hyperspectral image classification. For this purpose, a 3D fast learning block (depthwise separable convolution block and a fast convolution block) followed by a 2D convolutional neural network was introduced to extract spectralspatial features. Four datasets were used for the experiment purpose and the results showed that the proposed method achieved excellent result on both small and large training data compared with existing methods. The proposed method increased the overall accuracies by 2% on UP and KSC datasets while significantly reducing the training time on IP and KSC datasets, respectively. The proposed method increased all accuracies for at least 6% on IP, KSC and UP datasets when compared to some state‐of‐the‐art methods. Also, it reduced considerably the training and testing time on IP and KSC datasets when fast convolution block alone is involved.https://doi.org/10.1049/ipr2.12087Image recognitionComputer vision and image processing techniquesMachine learning (artificial intelligence)Neural nets |
| spellingShingle | Alou Diakite Gui Jiangsheng Fu Xiaping Hyperspectral image classification using 3D 2D CNN IET Image Processing Image recognition Computer vision and image processing techniques Machine learning (artificial intelligence) Neural nets |
| title | Hyperspectral image classification using 3D 2D CNN |
| title_full | Hyperspectral image classification using 3D 2D CNN |
| title_fullStr | Hyperspectral image classification using 3D 2D CNN |
| title_full_unstemmed | Hyperspectral image classification using 3D 2D CNN |
| title_short | Hyperspectral image classification using 3D 2D CNN |
| title_sort | hyperspectral image classification using 3d 2d cnn |
| topic | Image recognition Computer vision and image processing techniques Machine learning (artificial intelligence) Neural nets |
| url | https://doi.org/10.1049/ipr2.12087 |
| work_keys_str_mv | AT aloudiakite hyperspectralimageclassificationusing3d2dcnn AT guijiangsheng hyperspectralimageclassificationusing3d2dcnn AT fuxiaping hyperspectralimageclassificationusing3d2dcnn |