EANet: Edge-Aware Network for the Extraction of Buildings from Aerial Images
Deep learning methods have been used to extract buildings from remote sensing images and have achieved state-of-the-art performance. Most previous work has emphasized the multi-scale fusion of features or the enhancement of more receptive fields to achieve global features rather than focusing on low...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2020-07-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/12/13/2161 |
| _version_ | 1797563186172395520 |
|---|---|
| author | Guang Yang Qian Zhang Guixu Zhang |
| author_facet | Guang Yang Qian Zhang Guixu Zhang |
| author_sort | Guang Yang |
| collection | DOAJ |
| description | Deep learning methods have been used to extract buildings from remote sensing images and have achieved state-of-the-art performance. Most previous work has emphasized the multi-scale fusion of features or the enhancement of more receptive fields to achieve global features rather than focusing on low-level details such as the edges. In this work, we propose a novel end-to-end edge-aware network, the EANet, and an edge-aware loss for getting accurate buildings from aerial images. Specifically, the architecture is composed of image segmentation networks and edge perception networks that, respectively, take charge of building prediction and edge investigation. The International Society for Photogrammetry and Remote Sensing (ISPRS) Potsdam segmentation benchmark and the Wuhan University (WHU) building benchmark were used to evaluate our approach, which, respectively, was found to achieve 90.19% and 93.33% intersection-over-union and top performance without using additional datasets, data augmentation, and post-processing. The EANet is effective in extracting buildings from aerial images, which shows that the quality of image segmentation can be improved by focusing on edge details. |
| first_indexed | 2024-03-10T18:40:02Z |
| format | Article |
| id | doaj.art-dcdb7ffa9f254ec289c8ba781c113e94 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-10T18:40:02Z |
| publishDate | 2020-07-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-dcdb7ffa9f254ec289c8ba781c113e942023-11-20T05:59:53ZengMDPI AGRemote Sensing2072-42922020-07-011213216110.3390/rs12132161EANet: Edge-Aware Network for the Extraction of Buildings from Aerial ImagesGuang Yang0Qian Zhang1Guixu Zhang2The Shanghai Key Laboratory of Multidimensional Information Processing, East China Normal University, Shanghai 200241, ChinaThe Shanghai Key Laboratory of Multidimensional Information Processing, East China Normal University, Shanghai 200241, ChinaThe Shanghai Key Laboratory of Multidimensional Information Processing, East China Normal University, Shanghai 200241, ChinaDeep learning methods have been used to extract buildings from remote sensing images and have achieved state-of-the-art performance. Most previous work has emphasized the multi-scale fusion of features or the enhancement of more receptive fields to achieve global features rather than focusing on low-level details such as the edges. In this work, we propose a novel end-to-end edge-aware network, the EANet, and an edge-aware loss for getting accurate buildings from aerial images. Specifically, the architecture is composed of image segmentation networks and edge perception networks that, respectively, take charge of building prediction and edge investigation. The International Society for Photogrammetry and Remote Sensing (ISPRS) Potsdam segmentation benchmark and the Wuhan University (WHU) building benchmark were used to evaluate our approach, which, respectively, was found to achieve 90.19% and 93.33% intersection-over-union and top performance without using additional datasets, data augmentation, and post-processing. The EANet is effective in extracting buildings from aerial images, which shows that the quality of image segmentation can be improved by focusing on edge details.https://www.mdpi.com/2072-4292/12/13/2161semantic segmentationconvolutional neural networksbuilding extractionedgemulti-task learning |
| spellingShingle | Guang Yang Qian Zhang Guixu Zhang EANet: Edge-Aware Network for the Extraction of Buildings from Aerial Images Remote Sensing semantic segmentation convolutional neural networks building extraction edge multi-task learning |
| title | EANet: Edge-Aware Network for the Extraction of Buildings from Aerial Images |
| title_full | EANet: Edge-Aware Network for the Extraction of Buildings from Aerial Images |
| title_fullStr | EANet: Edge-Aware Network for the Extraction of Buildings from Aerial Images |
| title_full_unstemmed | EANet: Edge-Aware Network for the Extraction of Buildings from Aerial Images |
| title_short | EANet: Edge-Aware Network for the Extraction of Buildings from Aerial Images |
| title_sort | eanet edge aware network for the extraction of buildings from aerial images |
| topic | semantic segmentation convolutional neural networks building extraction edge multi-task learning |
| url | https://www.mdpi.com/2072-4292/12/13/2161 |
| work_keys_str_mv | AT guangyang eanetedgeawarenetworkfortheextractionofbuildingsfromaerialimages AT qianzhang eanetedgeawarenetworkfortheextractionofbuildingsfromaerialimages AT guixuzhang eanetedgeawarenetworkfortheextractionofbuildingsfromaerialimages |