Learn to Extract Building Outline from Misaligned Annotation through Nearest Feature Selector
For efficient building outline extraction, many algorithms, including unsupervised or supervised, have been proposed over the past decades. In recent years, due to the rapid development of the convolutional neural networks, especially fully convolutional networks, building extraction is treated as a...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2020-08-01
|
Series: | Remote Sensing |
Subjects: | |
Online Access: | https://www.mdpi.com/2072-4292/12/17/2722 |
id |
doaj-01f93d6b364c45b2866bf06d5b6945cb |
---|---|
record_format |
Article |
spelling |
doaj-01f93d6b364c45b2866bf06d5b6945cb2020-11-25T03:56:50ZengMDPI AGRemote Sensing2072-42922020-08-01122722272210.3390/rs12172722Learn to Extract Building Outline from Misaligned Annotation through Nearest Feature SelectorYuxuan Wang0Guangming Wu1Yimin Guo2Yifei Huang3Ryosuke Shibasaki4Center for Spatial Information Science, The University of Tokyo, Kashiwa 277-8568, JapanCenter for Spatial Information Science, The University of Tokyo, Kashiwa 277-8568, JapanCenter for Spatial Information Science, The University of Tokyo, Kashiwa 277-8568, JapanInstitute of Industrial Science, The University of Tokyo, Tokyo 153-8505, JapanCenter for Spatial Information Science, The University of Tokyo, Kashiwa 277-8568, JapanFor efficient building outline extraction, many algorithms, including unsupervised or supervised, have been proposed over the past decades. In recent years, due to the rapid development of the convolutional neural networks, especially fully convolutional networks, building extraction is treated as a semantic segmentation task that deals with the extremely biased positive pixels. The state-of-the-art methods, either through direct or indirect approaches, are mainly focused on better network design. The shifts and rotations, which are coarsely presented in manually created annotations, have long been ignored. Due to the limited number of positive samples, the misalignment will significantly reduce the correctness of pixel-to-pixel loss that might lead to a gradient explosion. To overcome this, we propose a nearest feature selector (NFS) to dynamically re-align the prediction and slightly misaligned annotations. The NFS can be seamlessly appended to existing loss functions and prevent misleading by the errors or misalignment of annotations. Experiments on a large scale aerial image dataset with centered buildings and corresponding building outlines indicate that the additional NFS brings higher performance when compared to existing naive loss functions. In the classic L1 loss, the addition of NFS gains increments of 8.8% of f1-score, 8.9% of kappa coefficient, and 9.8% of Jaccard index, respectively.https://www.mdpi.com/2072-4292/12/17/2722deep convolutional networksoutline extractionmisalignmentsnearest feature selector |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Yuxuan Wang Guangming Wu Yimin Guo Yifei Huang Ryosuke Shibasaki |
spellingShingle |
Yuxuan Wang Guangming Wu Yimin Guo Yifei Huang Ryosuke Shibasaki Learn to Extract Building Outline from Misaligned Annotation through Nearest Feature Selector Remote Sensing deep convolutional networks outline extraction misalignments nearest feature selector |
author_facet |
Yuxuan Wang Guangming Wu Yimin Guo Yifei Huang Ryosuke Shibasaki |
author_sort |
Yuxuan Wang |
title |
Learn to Extract Building Outline from Misaligned Annotation through Nearest Feature Selector |
title_short |
Learn to Extract Building Outline from Misaligned Annotation through Nearest Feature Selector |
title_full |
Learn to Extract Building Outline from Misaligned Annotation through Nearest Feature Selector |
title_fullStr |
Learn to Extract Building Outline from Misaligned Annotation through Nearest Feature Selector |
title_full_unstemmed |
Learn to Extract Building Outline from Misaligned Annotation through Nearest Feature Selector |
title_sort |
learn to extract building outline from misaligned annotation through nearest feature selector |
publisher |
MDPI AG |
series |
Remote Sensing |
issn |
2072-4292 |
publishDate |
2020-08-01 |
description |
For efficient building outline extraction, many algorithms, including unsupervised or supervised, have been proposed over the past decades. In recent years, due to the rapid development of the convolutional neural networks, especially fully convolutional networks, building extraction is treated as a semantic segmentation task that deals with the extremely biased positive pixels. The state-of-the-art methods, either through direct or indirect approaches, are mainly focused on better network design. The shifts and rotations, which are coarsely presented in manually created annotations, have long been ignored. Due to the limited number of positive samples, the misalignment will significantly reduce the correctness of pixel-to-pixel loss that might lead to a gradient explosion. To overcome this, we propose a nearest feature selector (NFS) to dynamically re-align the prediction and slightly misaligned annotations. The NFS can be seamlessly appended to existing loss functions and prevent misleading by the errors or misalignment of annotations. Experiments on a large scale aerial image dataset with centered buildings and corresponding building outlines indicate that the additional NFS brings higher performance when compared to existing naive loss functions. In the classic L1 loss, the addition of NFS gains increments of 8.8% of f1-score, 8.9% of kappa coefficient, and 9.8% of Jaccard index, respectively. |
topic |
deep convolutional networks outline extraction misalignments nearest feature selector |
url |
https://www.mdpi.com/2072-4292/12/17/2722 |
work_keys_str_mv |
AT yuxuanwang learntoextractbuildingoutlinefrommisalignedannotationthroughnearestfeatureselector AT guangmingwu learntoextractbuildingoutlinefrommisalignedannotationthroughnearestfeatureselector AT yiminguo learntoextractbuildingoutlinefrommisalignedannotationthroughnearestfeatureselector AT yifeihuang learntoextractbuildingoutlinefrommisalignedannotationthroughnearestfeatureselector AT ryosukeshibasaki learntoextractbuildingoutlinefrommisalignedannotationthroughnearestfeatureselector |
_version_ |
1724463504889479168 |