Leveraging Deep Neural Networks to Map Caribou Lichen in High-Resolution Satellite Images Based on a Small-Scale, Noisy UAV-Derived Map
Lichen is an important food source for caribou in Canada. Lichen mapping using remote sensing (RS) images could be a challenging task, however, as lichens generally appear in unevenly distributed, small patches, and could resemble surficial features. Moreover, collecting lichen labeled data (referen...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-07-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/13/14/2658 |
| _version_ | 1797526154162208768 |
|---|---|
| author | Shahab Jozdani Dongmei Chen Wenjun Chen Sylvain G. Leblanc Christian Prévost Julie Lovitt Liming He Brian A. Johnson |
| author_facet | Shahab Jozdani Dongmei Chen Wenjun Chen Sylvain G. Leblanc Christian Prévost Julie Lovitt Liming He Brian A. Johnson |
| author_sort | Shahab Jozdani |
| collection | DOAJ |
| description | Lichen is an important food source for caribou in Canada. Lichen mapping using remote sensing (RS) images could be a challenging task, however, as lichens generally appear in unevenly distributed, small patches, and could resemble surficial features. Moreover, collecting lichen labeled data (reference data) is expensive, which restricts the application of many robust supervised classification models that generally demand a large quantity of labeled data. The goal of this study was to investigate the potential of using a very-high-spatial resolution (1-cm) lichen map of a small sample site (e.g., generated based on a single UAV scene and using field data) to train a subsequent classifier to map caribou lichen over a much larger area (~0.04 km<sup>2</sup> vs. ~195 km<sup>2</sup>) and a lower spatial resolution image (in this case, a 50-cm WorldView-2 image). The limited labeled data from the sample site were also partially noisy due to spatial and temporal mismatching issues. For this, we deployed a recently proposed <i>Teacher-Student</i> semi-supervised learning (SSL) approach (based on U-Net and U-Net++ networks) involving unlabeled data to assist with improving the model performance. Our experiments showed that it was possible to scale-up the UAV-derived lichen map to the WorldView-2 scale with reasonable accuracy (overall accuracy of 85.28% and F1-socre of 84.38%) without collecting any samples directly in the WorldView-2 scene. We also found that our noisy labels were partially beneficial to the SSL robustness because they improved the false positive rate compared to the use of a cleaner training set directly collected within the same area in the WorldView-2 image. As a result, this research opens new insights into how current very high-resolution, small-scale caribou lichen maps can be used for generating more accurate large-scale caribou lichen maps from high-resolution satellite imagery. |
| first_indexed | 2024-03-10T09:26:08Z |
| format | Article |
| id | doaj.art-ecf5100189f945cdadf87f57afce481c |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-10T09:26:08Z |
| publishDate | 2021-07-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-ecf5100189f945cdadf87f57afce481c2023-11-22T04:50:31ZengMDPI AGRemote Sensing2072-42922021-07-011314265810.3390/rs13142658Leveraging Deep Neural Networks to Map Caribou Lichen in High-Resolution Satellite Images Based on a Small-Scale, Noisy UAV-Derived MapShahab Jozdani0Dongmei Chen1Wenjun Chen2Sylvain G. Leblanc3Christian Prévost4Julie Lovitt5Liming He6Brian A. Johnson7Department of Geography and Planning, Queen’s University, Kingston, ON K7L 3N6, CanadaDepartment of Geography and Planning, Queen’s University, Kingston, ON K7L 3N6, CanadaCanada Centre for Remote Sensing, Natural Resources Canada, 560 Rochester Street, Ottawa, ON K1S 5K2, CanadaCanada Centre for Remote Sensing, Natural Resources Canada, 560 Rochester Street, Ottawa, ON K1S 5K2, CanadaCanada Centre for Remote Sensing, Natural Resources Canada, 560 Rochester Street, Ottawa, ON K1S 5K2, CanadaCanada Centre for Remote Sensing, Natural Resources Canada, 560 Rochester Street, Ottawa, ON K1S 5K2, CanadaCanada Centre for Remote Sensing, Natural Resources Canada, 560 Rochester Street, Ottawa, ON K1S 5K2, CanadaNatural Resources and Ecosystem Services Area, Institute for Global Environmental Strategies, 2108-1 Kamiyamaguchi, Hayama, Kanagawa 240-0115, JapanLichen is an important food source for caribou in Canada. Lichen mapping using remote sensing (RS) images could be a challenging task, however, as lichens generally appear in unevenly distributed, small patches, and could resemble surficial features. Moreover, collecting lichen labeled data (reference data) is expensive, which restricts the application of many robust supervised classification models that generally demand a large quantity of labeled data. The goal of this study was to investigate the potential of using a very-high-spatial resolution (1-cm) lichen map of a small sample site (e.g., generated based on a single UAV scene and using field data) to train a subsequent classifier to map caribou lichen over a much larger area (~0.04 km<sup>2</sup> vs. ~195 km<sup>2</sup>) and a lower spatial resolution image (in this case, a 50-cm WorldView-2 image). The limited labeled data from the sample site were also partially noisy due to spatial and temporal mismatching issues. For this, we deployed a recently proposed <i>Teacher-Student</i> semi-supervised learning (SSL) approach (based on U-Net and U-Net++ networks) involving unlabeled data to assist with improving the model performance. Our experiments showed that it was possible to scale-up the UAV-derived lichen map to the WorldView-2 scale with reasonable accuracy (overall accuracy of 85.28% and F1-socre of 84.38%) without collecting any samples directly in the WorldView-2 scene. We also found that our noisy labels were partially beneficial to the SSL robustness because they improved the false positive rate compared to the use of a cleaner training set directly collected within the same area in the WorldView-2 image. As a result, this research opens new insights into how current very high-resolution, small-scale caribou lichen maps can be used for generating more accurate large-scale caribou lichen maps from high-resolution satellite imagery.https://www.mdpi.com/2072-4292/13/14/2658remote sensinglichen mappingdeep learningsemi-supervised learningteacher-student learningWorldView-2 |
| spellingShingle | Shahab Jozdani Dongmei Chen Wenjun Chen Sylvain G. Leblanc Christian Prévost Julie Lovitt Liming He Brian A. Johnson Leveraging Deep Neural Networks to Map Caribou Lichen in High-Resolution Satellite Images Based on a Small-Scale, Noisy UAV-Derived Map Remote Sensing remote sensing lichen mapping deep learning semi-supervised learning teacher-student learning WorldView-2 |
| title | Leveraging Deep Neural Networks to Map Caribou Lichen in High-Resolution Satellite Images Based on a Small-Scale, Noisy UAV-Derived Map |
| title_full | Leveraging Deep Neural Networks to Map Caribou Lichen in High-Resolution Satellite Images Based on a Small-Scale, Noisy UAV-Derived Map |
| title_fullStr | Leveraging Deep Neural Networks to Map Caribou Lichen in High-Resolution Satellite Images Based on a Small-Scale, Noisy UAV-Derived Map |
| title_full_unstemmed | Leveraging Deep Neural Networks to Map Caribou Lichen in High-Resolution Satellite Images Based on a Small-Scale, Noisy UAV-Derived Map |
| title_short | Leveraging Deep Neural Networks to Map Caribou Lichen in High-Resolution Satellite Images Based on a Small-Scale, Noisy UAV-Derived Map |
| title_sort | leveraging deep neural networks to map caribou lichen in high resolution satellite images based on a small scale noisy uav derived map |
| topic | remote sensing lichen mapping deep learning semi-supervised learning teacher-student learning WorldView-2 |
| url | https://www.mdpi.com/2072-4292/13/14/2658 |
| work_keys_str_mv | AT shahabjozdani leveragingdeepneuralnetworkstomapcariboulicheninhighresolutionsatelliteimagesbasedonasmallscalenoisyuavderivedmap AT dongmeichen leveragingdeepneuralnetworkstomapcariboulicheninhighresolutionsatelliteimagesbasedonasmallscalenoisyuavderivedmap AT wenjunchen leveragingdeepneuralnetworkstomapcariboulicheninhighresolutionsatelliteimagesbasedonasmallscalenoisyuavderivedmap AT sylvaingleblanc leveragingdeepneuralnetworkstomapcariboulicheninhighresolutionsatelliteimagesbasedonasmallscalenoisyuavderivedmap AT christianprevost leveragingdeepneuralnetworkstomapcariboulicheninhighresolutionsatelliteimagesbasedonasmallscalenoisyuavderivedmap AT julielovitt leveragingdeepneuralnetworkstomapcariboulicheninhighresolutionsatelliteimagesbasedonasmallscalenoisyuavderivedmap AT liminghe leveragingdeepneuralnetworkstomapcariboulicheninhighresolutionsatelliteimagesbasedonasmallscalenoisyuavderivedmap AT brianajohnson leveragingdeepneuralnetworkstomapcariboulicheninhighresolutionsatelliteimagesbasedonasmallscalenoisyuavderivedmap |