Semantic Boosting: Enhancing Deep Learning Based LULC Classification
The classification of land use and land cover (LULC) is a well-studied task within the domain of remote sensing and geographic information science. It traditionally relies on remotely sensed imagery and therefore models land cover classes with respect to their electromagnetic reflectances, aggregate...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-08-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/13/16/3197 |
| _version_ | 1797522129095229440 |
|---|---|
| author | Marvin Mc Cutchan Alexis J. Comber Ioannis Giannopoulos Manuela Canestrini |
| author_facet | Marvin Mc Cutchan Alexis J. Comber Ioannis Giannopoulos Manuela Canestrini |
| author_sort | Marvin Mc Cutchan |
| collection | DOAJ |
| description | The classification of land use and land cover (LULC) is a well-studied task within the domain of remote sensing and geographic information science. It traditionally relies on remotely sensed imagery and therefore models land cover classes with respect to their electromagnetic reflectances, aggregated in pixels. This paper introduces a methodology which enables the inclusion of geographical object semantics (from vector data) into the LULC classification procedure. As such, information on the types of geographic objects (e.g., <i>Shop</i>, <i>Church</i>, <i>Peak</i>, etc.) can improve LULC classification accuracy. In this paper, we demonstrate how semantics can be fused with imagery to classify LULC. Three experiments were performed to explore and highlight the impact and potential of semantics for this task. In each experiment CORINE LULC data was used as a ground truth and predicted using imagery from Sentinel-2 and semantics from LinkedGeoData using deep learning. Our results reveal that LULC can be classified from semantics only and that fusing semantics with imagery—Semantic Boosting—improved the classification with significantly higher LULC accuracies. The results show that some LULC classes are better predicted using only semantics, others with just imagery, and importantly much of the improvement was due to the ability to separate similar land use classes. A number of key considerations are discussed. |
| first_indexed | 2024-03-10T08:25:04Z |
| format | Article |
| id | doaj.art-139ef6e049f74c9da7d4351ce6ab2c99 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-10T08:25:04Z |
| publishDate | 2021-08-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-139ef6e049f74c9da7d4351ce6ab2c992023-11-22T09:33:37ZengMDPI AGRemote Sensing2072-42922021-08-011316319710.3390/rs13163197Semantic Boosting: Enhancing Deep Learning Based LULC ClassificationMarvin Mc Cutchan0Alexis J. Comber1Ioannis Giannopoulos2Manuela Canestrini3Department of Geodesy and Geoinformation, TU Wien, 1040 Vienna, AustriaLeeds Institute for Data Analytics, University of Leeds, Leeds LS2 9NL, UKDepartment of Geodesy and Geoinformation, TU Wien, 1040 Vienna, AustriaDepartment of Geodesy and Geoinformation, TU Wien, 1040 Vienna, AustriaThe classification of land use and land cover (LULC) is a well-studied task within the domain of remote sensing and geographic information science. It traditionally relies on remotely sensed imagery and therefore models land cover classes with respect to their electromagnetic reflectances, aggregated in pixels. This paper introduces a methodology which enables the inclusion of geographical object semantics (from vector data) into the LULC classification procedure. As such, information on the types of geographic objects (e.g., <i>Shop</i>, <i>Church</i>, <i>Peak</i>, etc.) can improve LULC classification accuracy. In this paper, we demonstrate how semantics can be fused with imagery to classify LULC. Three experiments were performed to explore and highlight the impact and potential of semantics for this task. In each experiment CORINE LULC data was used as a ground truth and predicted using imagery from Sentinel-2 and semantics from LinkedGeoData using deep learning. Our results reveal that LULC can be classified from semantics only and that fusing semantics with imagery—Semantic Boosting—improved the classification with significantly higher LULC accuracies. The results show that some LULC classes are better predicted using only semantics, others with just imagery, and importantly much of the improvement was due to the ability to separate similar land use classes. A number of key considerations are discussed.https://www.mdpi.com/2072-4292/13/16/3197land use and land cover classificationdeep learninggeospatial semanticsdata fusion |
| spellingShingle | Marvin Mc Cutchan Alexis J. Comber Ioannis Giannopoulos Manuela Canestrini Semantic Boosting: Enhancing Deep Learning Based LULC Classification Remote Sensing land use and land cover classification deep learning geospatial semantics data fusion |
| title | Semantic Boosting: Enhancing Deep Learning Based LULC Classification |
| title_full | Semantic Boosting: Enhancing Deep Learning Based LULC Classification |
| title_fullStr | Semantic Boosting: Enhancing Deep Learning Based LULC Classification |
| title_full_unstemmed | Semantic Boosting: Enhancing Deep Learning Based LULC Classification |
| title_short | Semantic Boosting: Enhancing Deep Learning Based LULC Classification |
| title_sort | semantic boosting enhancing deep learning based lulc classification |
| topic | land use and land cover classification deep learning geospatial semantics data fusion |
| url | https://www.mdpi.com/2072-4292/13/16/3197 |
| work_keys_str_mv | AT marvinmccutchan semanticboostingenhancingdeeplearningbasedlulcclassification AT alexisjcomber semanticboostingenhancingdeeplearningbasedlulcclassification AT ioannisgiannopoulos semanticboostingenhancingdeeplearningbasedlulcclassification AT manuelacanestrini semanticboostingenhancingdeeplearningbasedlulcclassification |