SELF-TRAINING FOR SEMI-SUPERVISED DEEP CONTOUR DETECTION OF SURFACE WATER
Contour detection is better for monitoring dynamic and long-term changes to surface water bodies. For that purpose, we present a semi-automated method for collecting and labeling water contours from Landsat-8 and Sentinel-2 images. Due to the need for human inspection, the method has thus far genera...
Main Authors: | , |
---|---|
Format: | Article |
Language: | English |
Published: |
Copernicus Publications
2022-05-01
|
Series: | The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
Online Access: | https://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XLIII-B3-2022/1393/2022/isprs-archives-XLIII-B3-2022-1393-2022.pdf |
_version_ | 1811233070526758912 |
---|---|
author | A. Alsamman M. B. Syed |
author_facet | A. Alsamman M. B. Syed |
author_sort | A. Alsamman |
collection | DOAJ |
description | Contour detection is better for monitoring dynamic and long-term changes to surface water bodies. For that purpose, we present a semi-automated method for collecting and labeling water contours from Landsat-8 and Sentinel-2 images. Due to the need for human inspection, the method has thus far generated 14K labeled images from more than 1.5M images. Given the cost of data labeling, we propose a deep semi-supervised self-learning system performed in two training stages, known as teacher-student. The teacher is trained on the accurate human-labeled data, then used to pseudo label the remaining unlabeled data. The student is trained on both human-labeled and machine pseudo-labeled data. For both teacher and student, we use a uniquely designed multiscale UNet classifier that uses fewer parameters and is more accurate than other state-of-the-art classifiers. Random augmentations are used to “noise” the student model and improve its generalization, and normalization schemes are used to blend the human-labeled loss with the machine-labeled loss. Comparisons to existing water body detection classifiers and segmentation classifiers show the superiority of our proposed system in detecting water contours. |
first_indexed | 2024-04-12T11:15:21Z |
format | Article |
id | doaj.art-0ea6cf9cef39421787f841c3db0e11ed |
institution | Directory Open Access Journal |
issn | 1682-1750 2194-9034 |
language | English |
last_indexed | 2024-04-12T11:15:21Z |
publishDate | 2022-05-01 |
publisher | Copernicus Publications |
record_format | Article |
series | The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
spelling | doaj.art-0ea6cf9cef39421787f841c3db0e11ed2022-12-22T03:35:31ZengCopernicus PublicationsThe International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences1682-17502194-90342022-05-01XLIII-B3-20221393139810.5194/isprs-archives-XLIII-B3-2022-1393-2022SELF-TRAINING FOR SEMI-SUPERVISED DEEP CONTOUR DETECTION OF SURFACE WATERA. Alsamman0M. B. Syed1Department of Electrical and Computer Engineering, University of New Orleans, USADepartment of Electrical and Computer Engineering, University of New Orleans, USAContour detection is better for monitoring dynamic and long-term changes to surface water bodies. For that purpose, we present a semi-automated method for collecting and labeling water contours from Landsat-8 and Sentinel-2 images. Due to the need for human inspection, the method has thus far generated 14K labeled images from more than 1.5M images. Given the cost of data labeling, we propose a deep semi-supervised self-learning system performed in two training stages, known as teacher-student. The teacher is trained on the accurate human-labeled data, then used to pseudo label the remaining unlabeled data. The student is trained on both human-labeled and machine pseudo-labeled data. For both teacher and student, we use a uniquely designed multiscale UNet classifier that uses fewer parameters and is more accurate than other state-of-the-art classifiers. Random augmentations are used to “noise” the student model and improve its generalization, and normalization schemes are used to blend the human-labeled loss with the machine-labeled loss. Comparisons to existing water body detection classifiers and segmentation classifiers show the superiority of our proposed system in detecting water contours.https://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XLIII-B3-2022/1393/2022/isprs-archives-XLIII-B3-2022-1393-2022.pdf |
spellingShingle | A. Alsamman M. B. Syed SELF-TRAINING FOR SEMI-SUPERVISED DEEP CONTOUR DETECTION OF SURFACE WATER The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
title | SELF-TRAINING FOR SEMI-SUPERVISED DEEP CONTOUR DETECTION OF SURFACE WATER |
title_full | SELF-TRAINING FOR SEMI-SUPERVISED DEEP CONTOUR DETECTION OF SURFACE WATER |
title_fullStr | SELF-TRAINING FOR SEMI-SUPERVISED DEEP CONTOUR DETECTION OF SURFACE WATER |
title_full_unstemmed | SELF-TRAINING FOR SEMI-SUPERVISED DEEP CONTOUR DETECTION OF SURFACE WATER |
title_short | SELF-TRAINING FOR SEMI-SUPERVISED DEEP CONTOUR DETECTION OF SURFACE WATER |
title_sort | self training for semi supervised deep contour detection of surface water |
url | https://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XLIII-B3-2022/1393/2022/isprs-archives-XLIII-B3-2022-1393-2022.pdf |
work_keys_str_mv | AT aalsamman selftrainingforsemisuperviseddeepcontourdetectionofsurfacewater AT mbsyed selftrainingforsemisuperviseddeepcontourdetectionofsurfacewater |