Generating Training Data for Deep Learning-Based Segmentation Algorithms by Projecting Existing Labels onto Additional Aerial Images
Highly accurate manually-generated labels in aerial and satellite images are used for the training of deep learning-based segmentation algorithms and should be available in large numbers and cover many different scenarios to increase the accuracy and generalization capability of the underlying model...
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Copernicus Publications
2025-05-01
|
| Series: | The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| Online Access: | https://isprs-archives.copernicus.org/articles/XLVIII-M-6-2025/189/2025/isprs-archives-XLVIII-M-6-2025-189-2025.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849735883998101504 |
|---|---|
| author | F. Kurz N. Merkle C. Henry R. Bahmanyar F. Rauch J. Hellekes V. Gstaiger D. Rosenbaum P. Reinartz |
| author_facet | F. Kurz N. Merkle C. Henry R. Bahmanyar F. Rauch J. Hellekes V. Gstaiger D. Rosenbaum P. Reinartz |
| author_sort | F. Kurz |
| collection | DOAJ |
| description | Highly accurate manually-generated labels in aerial and satellite images are used for the training of deep learning-based segmentation algorithms and should be available in large numbers and cover many different scenarios to increase the accuracy and generalization capability of the underlying models. Existing labels can be efficiently reused by photogrammetric projections onto additional overlapping aerial or satellite images, enabling great variability in the appearance of the scenes based on differences in viewing angles and environmental conditions. In this work, we investigate whether the additionally generated training data can effectively lead to an increase in prediction accuracy. To this end, we collected aerial images overlapping with the already annotated Traffic Infrastructure and Surroundings (TIAS) dataset, taken from a large-scale historical database spanning 2011 to 2024, and generated new training data by means of photogrammetric projections of existing labels onto these additional images. Training a Dense-U-Net model on the whole TIAS dataset or a part therefore, with and without additional projected labels, showed that this technique could be beneficial to improve the performance of a model if only a small amount of annotations is available comparatively to a large amount of overlapping aerial images. |
| format | Article |
| id | doaj-art-5b845ae3faa3404aa1b7e6fa9e63d09e |
| institution | DOAJ |
| issn | 1682-1750 2194-9034 |
| language | English |
| publishDate | 2025-05-01 |
| publisher | Copernicus Publications |
| record_format | Article |
| series | The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| spelling | doaj-art-5b845ae3faa3404aa1b7e6fa9e63d09e2025-08-20T03:07:26ZengCopernicus PublicationsThe International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences1682-17502194-90342025-05-01XLVIII-M-6-202518919510.5194/isprs-archives-XLVIII-M-6-2025-189-2025Generating Training Data for Deep Learning-Based Segmentation Algorithms by Projecting Existing Labels onto Additional Aerial ImagesF. Kurz0N. Merkle1C. Henry2R. Bahmanyar3F. Rauch4J. Hellekes5V. Gstaiger6D. Rosenbaum7P. Reinartz8DLR, Earth Observation Center, 82234 Wessling, GermanyDLR, Earth Observation Center, 82234 Wessling, GermanyDLR, Earth Observation Center, 82234 Wessling, GermanyDLR, Earth Observation Center, 82234 Wessling, GermanyDLR, Earth Observation Center, 82234 Wessling, GermanyDLR, Earth Observation Center, 82234 Wessling, GermanyDLR, Earth Observation Center, 82234 Wessling, GermanyDLR, Earth Observation Center, 82234 Wessling, GermanyDLR, Earth Observation Center, 82234 Wessling, GermanyHighly accurate manually-generated labels in aerial and satellite images are used for the training of deep learning-based segmentation algorithms and should be available in large numbers and cover many different scenarios to increase the accuracy and generalization capability of the underlying models. Existing labels can be efficiently reused by photogrammetric projections onto additional overlapping aerial or satellite images, enabling great variability in the appearance of the scenes based on differences in viewing angles and environmental conditions. In this work, we investigate whether the additionally generated training data can effectively lead to an increase in prediction accuracy. To this end, we collected aerial images overlapping with the already annotated Traffic Infrastructure and Surroundings (TIAS) dataset, taken from a large-scale historical database spanning 2011 to 2024, and generated new training data by means of photogrammetric projections of existing labels onto these additional images. Training a Dense-U-Net model on the whole TIAS dataset or a part therefore, with and without additional projected labels, showed that this technique could be beneficial to improve the performance of a model if only a small amount of annotations is available comparatively to a large amount of overlapping aerial images.https://isprs-archives.copernicus.org/articles/XLVIII-M-6-2025/189/2025/isprs-archives-XLVIII-M-6-2025-189-2025.pdf |
| spellingShingle | F. Kurz N. Merkle C. Henry R. Bahmanyar F. Rauch J. Hellekes V. Gstaiger D. Rosenbaum P. Reinartz Generating Training Data for Deep Learning-Based Segmentation Algorithms by Projecting Existing Labels onto Additional Aerial Images The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| title | Generating Training Data for Deep Learning-Based Segmentation Algorithms by Projecting Existing Labels onto Additional Aerial Images |
| title_full | Generating Training Data for Deep Learning-Based Segmentation Algorithms by Projecting Existing Labels onto Additional Aerial Images |
| title_fullStr | Generating Training Data for Deep Learning-Based Segmentation Algorithms by Projecting Existing Labels onto Additional Aerial Images |
| title_full_unstemmed | Generating Training Data for Deep Learning-Based Segmentation Algorithms by Projecting Existing Labels onto Additional Aerial Images |
| title_short | Generating Training Data for Deep Learning-Based Segmentation Algorithms by Projecting Existing Labels onto Additional Aerial Images |
| title_sort | generating training data for deep learning based segmentation algorithms by projecting existing labels onto additional aerial images |
| url | https://isprs-archives.copernicus.org/articles/XLVIII-M-6-2025/189/2025/isprs-archives-XLVIII-M-6-2025-189-2025.pdf |
| work_keys_str_mv | AT fkurz generatingtrainingdatafordeeplearningbasedsegmentationalgorithmsbyprojectingexistinglabelsontoadditionalaerialimages AT nmerkle generatingtrainingdatafordeeplearningbasedsegmentationalgorithmsbyprojectingexistinglabelsontoadditionalaerialimages AT chenry generatingtrainingdatafordeeplearningbasedsegmentationalgorithmsbyprojectingexistinglabelsontoadditionalaerialimages AT rbahmanyar generatingtrainingdatafordeeplearningbasedsegmentationalgorithmsbyprojectingexistinglabelsontoadditionalaerialimages AT frauch generatingtrainingdatafordeeplearningbasedsegmentationalgorithmsbyprojectingexistinglabelsontoadditionalaerialimages AT jhellekes generatingtrainingdatafordeeplearningbasedsegmentationalgorithmsbyprojectingexistinglabelsontoadditionalaerialimages AT vgstaiger generatingtrainingdatafordeeplearningbasedsegmentationalgorithmsbyprojectingexistinglabelsontoadditionalaerialimages AT drosenbaum generatingtrainingdatafordeeplearningbasedsegmentationalgorithmsbyprojectingexistinglabelsontoadditionalaerialimages AT preinartz generatingtrainingdatafordeeplearningbasedsegmentationalgorithmsbyprojectingexistinglabelsontoadditionalaerialimages |