View in EDS

Learning Aerial Image Segmentation From Online Maps.

Saved in:
Bibliographic Details
Title: Learning Aerial Image Segmentation From Online Maps.
Authors: Kaiser, Pascal1, Wegner, Jan Dirk1, Lucchi, Aurelien1, Jaggi, Martin1, Hofmann, Thomas1, Schindler, Konrad1
Source: IEEE Transactions on Geoscience & Remote Sensing. Nov2017, Vol. 55 Issue 11, p6054-6068. 15p.
Subjects: Image segmentation, High resolution imaging, Artificial neural networks, Semantic computing, Mathematical convolutions
Abstract: This paper deals with semantic segmentation of high-resolution (aerial) images where a semantic class label is assigned to each pixel via supervised classification as a basis for automatic map generation. Recently, deep convolutional neural networks (CNNs) have shown impressive performance and have quickly become the de-facto standard for semantic segmentation, with the added benefit that task-specific feature design is no longer necessary. However, a major downside of deep learning methods is that they are extremely data hungry, thus aggravating the perennial bottleneck of supervised classification, to obtain enough annotated training data. On the other hand, it has been observed that they are rather robust against noise in the training labels. This opens up the intriguing possibility to avoid annotating huge amounts of training data, and instead train the classifier from existing legacy data or crowd-sourced maps that can exhibit high levels of noise. The question addressed in this paper is: can training with large-scale publicly available labels replace a substantial part of the manual labeling effort and still achieve sufficient performance? Such data will inevitably contain a significant portion of errors, but in return virtually unlimited quantities of it are available in larger parts of the world. We adapt a state-of-the-art CNN architecture for semantic segmentation of buildings and roads in aerial images, and compare its performance when using different training data sets, ranging from manually labeled pixel-accurate ground truth of the same city to automatic training data derived from OpenStreetMap data from distant locations. We report our results that indicate that satisfying performance can be obtained with significantly less manual annotation effort, by exploiting noisy large-scale training data. [ABSTRACT FROM PUBLISHER]
Copyright of IEEE Transactions on Geoscience & Remote Sensing is the property of IEEE and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
FullText Text:
  Availability: 0
Header DbId: egs
DbLabel: Engineering Source
An: 125952123
AccessLevel: 6
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Learning Aerial Image Segmentation From Online Maps.
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Kaiser%2C+Pascal%22">Kaiser, Pascal</searchLink><relatesTo>1</relatesTo><br /><searchLink fieldCode="AR" term="%22Wegner%2C+Jan+Dirk%22">Wegner, Jan Dirk</searchLink><relatesTo>1</relatesTo><br /><searchLink fieldCode="AR" term="%22Lucchi%2C+Aurelien%22">Lucchi, Aurelien</searchLink><relatesTo>1</relatesTo><br /><searchLink fieldCode="AR" term="%22Jaggi%2C+Martin%22">Jaggi, Martin</searchLink><relatesTo>1</relatesTo><br /><searchLink fieldCode="AR" term="%22Hofmann%2C+Thomas%22">Hofmann, Thomas</searchLink><relatesTo>1</relatesTo><br /><searchLink fieldCode="AR" term="%22Schindler%2C+Konrad%22">Schindler, Konrad</searchLink><relatesTo>1</relatesTo>
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="JN" term="%22IEEE+Transactions+on+Geoscience+%26+Remote+Sensing%22">IEEE Transactions on Geoscience & Remote Sensing</searchLink>. Nov2017, Vol. 55 Issue 11, p6054-6068. 15p.
– Name: Subject
  Label: Subjects
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Image+segmentation%22">Image segmentation</searchLink><br /><searchLink fieldCode="DE" term="%22High+resolution+imaging%22">High resolution imaging</searchLink><br /><searchLink fieldCode="DE" term="%22Artificial+neural+networks%22">Artificial neural networks</searchLink><br /><searchLink fieldCode="DE" term="%22Semantic+computing%22">Semantic computing</searchLink><br /><searchLink fieldCode="DE" term="%22Mathematical+convolutions%22">Mathematical convolutions</searchLink>
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: This paper deals with semantic segmentation of high-resolution (aerial) images where a semantic class label is assigned to each pixel via supervised classification as a basis for automatic map generation. Recently, deep convolutional neural networks (CNNs) have shown impressive performance and have quickly become the de-facto standard for semantic segmentation, with the added benefit that task-specific feature design is no longer necessary. However, a major downside of deep learning methods is that they are extremely data hungry, thus aggravating the perennial bottleneck of supervised classification, to obtain enough annotated training data. On the other hand, it has been observed that they are rather robust against noise in the training labels. This opens up the intriguing possibility to avoid annotating huge amounts of training data, and instead train the classifier from existing legacy data or crowd-sourced maps that can exhibit high levels of noise. The question addressed in this paper is: can training with large-scale publicly available labels replace a substantial part of the manual labeling effort and still achieve sufficient performance? Such data will inevitably contain a significant portion of errors, but in return virtually unlimited quantities of it are available in larger parts of the world. We adapt a state-of-the-art CNN architecture for semantic segmentation of buildings and roads in aerial images, and compare its performance when using different training data sets, ranging from manually labeled pixel-accurate ground truth of the same city to automatic training data derived from OpenStreetMap data from distant locations. We report our results that indicate that satisfying performance can be obtained with significantly less manual annotation effort, by exploiting noisy large-scale training data. [ABSTRACT FROM PUBLISHER]
– Name: AbstractSuppliedCopyright
  Label:
  Group: Ab
  Data: <i>Copyright of IEEE Transactions on Geoscience & Remote Sensing is the property of IEEE and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract.</i> (Copyright applies to all Abstracts.)
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=egs&AN=125952123
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1109/TGRS.2017.2719738
    Languages:
      – Code: eng
        Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 15
        StartPage: 6054
    Subjects:
      – SubjectFull: Image segmentation
        Type: general
      – SubjectFull: High resolution imaging
        Type: general
      – SubjectFull: Artificial neural networks
        Type: general
      – SubjectFull: Semantic computing
        Type: general
      – SubjectFull: Mathematical convolutions
        Type: general
    Titles:
      – TitleFull: Learning Aerial Image Segmentation From Online Maps.
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Kaiser, Pascal
      – PersonEntity:
          Name:
            NameFull: Wegner, Jan Dirk
      – PersonEntity:
          Name:
            NameFull: Lucchi, Aurelien
      – PersonEntity:
          Name:
            NameFull: Jaggi, Martin
      – PersonEntity:
          Name:
            NameFull: Hofmann, Thomas
      – PersonEntity:
          Name:
            NameFull: Schindler, Konrad
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 11
              Text: Nov2017
              Type: published
              Y: 2017
          Identifiers:
            – Type: issn-print
              Value: 01962892
          Numbering:
            – Type: volume
              Value: 55
            – Type: issue
              Value: 11
          Titles:
            – TitleFull: IEEE Transactions on Geoscience & Remote Sensing
              Type: main
ResultId 1