Building outline delineation: From very high resolution remote sensing imagery to polygons with an improved end-to-end learning framework

W. Zhao*, I. Ivanov, C. Persello, A. Stein

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

7 Citations (Scopus)
214 Downloads (Pure)

Abstract

Deep learning methods based on Fully convolution networks (FCNs) have shown an impressive progress in building outline delineation from very high resolution (VHR) remote sensing (RS) imagery. Common issues still exist in extracting precise building shapes and outlines, often resulting in irregular edges and over smoothed corners. In this paper, we use PolyMapper, a recently introduced deep-learning framework that is able to predict object outlines in a vector representation directly. We have introduced two main modifications to this baseline method. First, we introduce EffcientNet as backbone feature encoder to our network, which uses compound coefficient to scale up all dimensions of depth/width/resolution uniformly, to improve the processing speed with fewer parameters. Second, we integrate a boundary refinement block (BRB) to strengthen the boundary feature learning and to further improve the accuracy of corner prediction. The results demonstrate that the end-to-end learnable model is capable of delineating polygons of building outlines that closely approximate the structure of reference labels. Experiments on the crowdAI building instance segmentation datasets show that our model outperforms PolyMapper in all COCO metrics, for instance showing a 0.13 higher mean Average Precision (AP) value and a 0.60 higher mean Average Recall value. Also qualitative results show that our method segments building instances of various shapes more accurately.

Original languageEnglish
Title of host publicationXXIV ISPRS Congress, Commission II 2020
EditorsN. Paparoditis, C. Mallet, F. Lafarge, F. Remondino, I. Toschi, T. Fuse
PublisherInternational Society for Photogrammetry and Remote Sensing (ISPRS)
Pages731-735
Number of pages5
DOIs
Publication statusPublished - 6 Aug 2020
EventXXIVth ISPRS Congress 2020 - Virtual Event, Nice, France
Duration: 4 Jul 202010 Jul 2020
Conference number: 24
http://www.isprs2020-nice.com

Publication series

NameInternational Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives
PublisherCopernicus
Volume43-B2
ISSN (Print)1682-1750

Conference

ConferenceXXIVth ISPRS Congress 2020
Abbreviated titleISPRS 2020
Country/TerritoryFrance
CityNice
Period4/07/2010/07/20
Internet address

Keywords

  • Building outline delineation
  • Convolutional Neural Networks (CNN)
  • Polygon prediction
  • Recurrent neural networks

Fingerprint

Dive into the research topics of 'Building outline delineation: From very high resolution remote sensing imagery to polygons with an improved end-to-end learning framework'. Together they form a unique fingerprint.

Cite this