Training Methods of Multi-Label Prediction Classifiers for Hyperspectral Remote Sensing Images
Hyperspectral remote sensing images, with their amalgamation of spectral richness and geometric precision, encapsulate intricate, non-linear information that poses significant challenges to traditional machine learning methodologies. Deep learning techniques, recognised for their superior representa...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-12-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/15/24/5656 |
| _version_ | 1797379520918978560 |
|---|---|
| author | Salma Haidar José Oramas |
| author_facet | Salma Haidar José Oramas |
| author_sort | Salma Haidar |
| collection | DOAJ |
| description | Hyperspectral remote sensing images, with their amalgamation of spectral richness and geometric precision, encapsulate intricate, non-linear information that poses significant challenges to traditional machine learning methodologies. Deep learning techniques, recognised for their superior representation learning capabilities, exhibit enhanced proficiency in managing such intricate data. In this study, we introduce a novel approach in hyperspectral image analysis focusing on multi-label, patch-level classification, as opposed to applications in the literature concentrating predominantly on single-label, pixel-level classification for hyperspectral remote sensing images. The proposed model comprises a two-component deep learning network and employs patches of hyperspectral remote sensing scenes with reduced spatial dimensions yet with a complete spectral depth derived from the original scene. Additionally, this work explores three distinct training schemes for our network: <i>Iterative</i>, <i>Joint</i>, and <i>Cascade</i>. Empirical evidence suggests the <i>Joint</i> approach as the optimal strategy, but it requires an extensive search to ascertain the optimal weight combination of the loss constituents. The <i>Iterative</i> scheme facilitates feature sharing between the network components from the early phases of training and demonstrates superior performance with complex, multi-labelled data. Subsequent analysis reveals that models with varying architectures, when trained on patches derived and annotated per our proposed single-label sampling procedure, exhibit commendable performance. |
| first_indexed | 2024-03-08T20:24:27Z |
| format | Article |
| id | doaj.art-ca94abb63d834347bea50d70c452b53a |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-08T20:24:27Z |
| publishDate | 2023-12-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-ca94abb63d834347bea50d70c452b53a2023-12-22T14:38:53ZengMDPI AGRemote Sensing2072-42922023-12-011524565610.3390/rs15245656Training Methods of Multi-Label Prediction Classifiers for Hyperspectral Remote Sensing ImagesSalma Haidar0José Oramas1Department of Computer Science, University of Antwerp, imec-IDLab, Sint-Pietersvliet 7, 2000 Antwerpen, BelgiumDepartment of Computer Science, University of Antwerp, imec-IDLab, Sint-Pietersvliet 7, 2000 Antwerpen, BelgiumHyperspectral remote sensing images, with their amalgamation of spectral richness and geometric precision, encapsulate intricate, non-linear information that poses significant challenges to traditional machine learning methodologies. Deep learning techniques, recognised for their superior representation learning capabilities, exhibit enhanced proficiency in managing such intricate data. In this study, we introduce a novel approach in hyperspectral image analysis focusing on multi-label, patch-level classification, as opposed to applications in the literature concentrating predominantly on single-label, pixel-level classification for hyperspectral remote sensing images. The proposed model comprises a two-component deep learning network and employs patches of hyperspectral remote sensing scenes with reduced spatial dimensions yet with a complete spectral depth derived from the original scene. Additionally, this work explores three distinct training schemes for our network: <i>Iterative</i>, <i>Joint</i>, and <i>Cascade</i>. Empirical evidence suggests the <i>Joint</i> approach as the optimal strategy, but it requires an extensive search to ascertain the optimal weight combination of the loss constituents. The <i>Iterative</i> scheme facilitates feature sharing between the network components from the early phases of training and demonstrates superior performance with complex, multi-labelled data. Subsequent analysis reveals that models with varying architectures, when trained on patches derived and annotated per our proposed single-label sampling procedure, exhibit commendable performance.https://www.mdpi.com/2072-4292/15/24/5656hyperspectral imagingcomputer visiondeep learningmulti-label classificationtwo-component neural networkdeep auto-encoder |
| spellingShingle | Salma Haidar José Oramas Training Methods of Multi-Label Prediction Classifiers for Hyperspectral Remote Sensing Images Remote Sensing hyperspectral imaging computer vision deep learning multi-label classification two-component neural network deep auto-encoder |
| title | Training Methods of Multi-Label Prediction Classifiers for Hyperspectral Remote Sensing Images |
| title_full | Training Methods of Multi-Label Prediction Classifiers for Hyperspectral Remote Sensing Images |
| title_fullStr | Training Methods of Multi-Label Prediction Classifiers for Hyperspectral Remote Sensing Images |
| title_full_unstemmed | Training Methods of Multi-Label Prediction Classifiers for Hyperspectral Remote Sensing Images |
| title_short | Training Methods of Multi-Label Prediction Classifiers for Hyperspectral Remote Sensing Images |
| title_sort | training methods of multi label prediction classifiers for hyperspectral remote sensing images |
| topic | hyperspectral imaging computer vision deep learning multi-label classification two-component neural network deep auto-encoder |
| url | https://www.mdpi.com/2072-4292/15/24/5656 |
| work_keys_str_mv | AT salmahaidar trainingmethodsofmultilabelpredictionclassifiersforhyperspectralremotesensingimages AT joseoramas trainingmethodsofmultilabelpredictionclassifiersforhyperspectralremotesensingimages |