Learning Deep Structured Active Contours End-to-End

Diego Marcos, Devis Tuia, Benjamin Kellenberger, Lisa Zhang, Min Bai, Renjie Liao, Raquel Urtasun

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

14 Citations (Scopus)

Abstract

The world is covered with millions of buildings, and precisely knowing each instance's position and extents is vital to a multitude of applications. Recently, automated building footprint segmentation models have shown superior detection accuracy thanks to the usage of Convolutional Neural Networks (CNN). However, even the latest evolutions struggle to precisely delineating borders, which often leads to geometric distortions and inadvertent fusion of adjacent building instances. We propose to overcome this issue by exploiting the distinct geometric properties of buildings. To this end, we present Deep Structured Active Contours (DSAC), a novel framework that integrates priors and constraints into the segmentation process, such as continuous boundaries, smooth edges, and sharp corners. To do so, DSAC employs Active Contour Models (ACM), a family of constraint- and prior-based polygonal models. We learn ACM parameterizations per instance using a CNN, and show how to incorporate all components in a structured output model, making DSAC trainable end-to-end. We evaluate DSAC on three challenging building instance segmentation datasets, where it compares favorably against state-of-the-art. Code will be made available on https://github.com/dmarcosg/DSAC.

Original languageEnglish
Title of host publicationProceedings - 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2018
PublisherIEEE computer society
Pages8877-8885
ISBN (Electronic)9781538664209
DOIs
Publication statusPublished - 14 Dec 2018
Event31st Meeting of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2018 - Salt Lake City, United States
Duration: 18 Jun 201822 Jun 2018

Conference

Conference31st Meeting of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2018
CountryUnited States
CitySalt Lake City
Period18/06/1822/06/18

Fingerprint

Neural networks
Parameterization
Fusion reactions
Deep learning

Cite this

Marcos, D., Tuia, D., Kellenberger, B., Zhang, L., Bai, M., Liao, R., & Urtasun, R. (2018). Learning Deep Structured Active Contours End-to-End. In Proceedings - 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2018 (pp. 8877-8885). [8579023] IEEE computer society. https://doi.org/10.1109/CVPR.2018.00925
Marcos, Diego ; Tuia, Devis ; Kellenberger, Benjamin ; Zhang, Lisa ; Bai, Min ; Liao, Renjie ; Urtasun, Raquel. / Learning Deep Structured Active Contours End-to-End. Proceedings - 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2018. IEEE computer society, 2018. pp. 8877-8885
@inproceedings{d65001f7dfe8400695c73406e85482c2,
title = "Learning Deep Structured Active Contours End-to-End",
abstract = "The world is covered with millions of buildings, and precisely knowing each instance's position and extents is vital to a multitude of applications. Recently, automated building footprint segmentation models have shown superior detection accuracy thanks to the usage of Convolutional Neural Networks (CNN). However, even the latest evolutions struggle to precisely delineating borders, which often leads to geometric distortions and inadvertent fusion of adjacent building instances. We propose to overcome this issue by exploiting the distinct geometric properties of buildings. To this end, we present Deep Structured Active Contours (DSAC), a novel framework that integrates priors and constraints into the segmentation process, such as continuous boundaries, smooth edges, and sharp corners. To do so, DSAC employs Active Contour Models (ACM), a family of constraint- and prior-based polygonal models. We learn ACM parameterizations per instance using a CNN, and show how to incorporate all components in a structured output model, making DSAC trainable end-to-end. We evaluate DSAC on three challenging building instance segmentation datasets, where it compares favorably against state-of-the-art. Code will be made available on https://github.com/dmarcosg/DSAC.",
author = "Diego Marcos and Devis Tuia and Benjamin Kellenberger and Lisa Zhang and Min Bai and Renjie Liao and Raquel Urtasun",
year = "2018",
month = "12",
day = "14",
doi = "10.1109/CVPR.2018.00925",
language = "English",
pages = "8877--8885",
booktitle = "Proceedings - 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2018",
publisher = "IEEE computer society",

}

Marcos, D, Tuia, D, Kellenberger, B, Zhang, L, Bai, M, Liao, R & Urtasun, R 2018, Learning Deep Structured Active Contours End-to-End. in Proceedings - 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2018., 8579023, IEEE computer society, pp. 8877-8885, 31st Meeting of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2018, Salt Lake City, United States, 18/06/18. https://doi.org/10.1109/CVPR.2018.00925

Learning Deep Structured Active Contours End-to-End. / Marcos, Diego; Tuia, Devis; Kellenberger, Benjamin; Zhang, Lisa; Bai, Min; Liao, Renjie; Urtasun, Raquel.

Proceedings - 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2018. IEEE computer society, 2018. p. 8877-8885 8579023.

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

TY - GEN

T1 - Learning Deep Structured Active Contours End-to-End

AU - Marcos, Diego

AU - Tuia, Devis

AU - Kellenberger, Benjamin

AU - Zhang, Lisa

AU - Bai, Min

AU - Liao, Renjie

AU - Urtasun, Raquel

PY - 2018/12/14

Y1 - 2018/12/14

N2 - The world is covered with millions of buildings, and precisely knowing each instance's position and extents is vital to a multitude of applications. Recently, automated building footprint segmentation models have shown superior detection accuracy thanks to the usage of Convolutional Neural Networks (CNN). However, even the latest evolutions struggle to precisely delineating borders, which often leads to geometric distortions and inadvertent fusion of adjacent building instances. We propose to overcome this issue by exploiting the distinct geometric properties of buildings. To this end, we present Deep Structured Active Contours (DSAC), a novel framework that integrates priors and constraints into the segmentation process, such as continuous boundaries, smooth edges, and sharp corners. To do so, DSAC employs Active Contour Models (ACM), a family of constraint- and prior-based polygonal models. We learn ACM parameterizations per instance using a CNN, and show how to incorporate all components in a structured output model, making DSAC trainable end-to-end. We evaluate DSAC on three challenging building instance segmentation datasets, where it compares favorably against state-of-the-art. Code will be made available on https://github.com/dmarcosg/DSAC.

AB - The world is covered with millions of buildings, and precisely knowing each instance's position and extents is vital to a multitude of applications. Recently, automated building footprint segmentation models have shown superior detection accuracy thanks to the usage of Convolutional Neural Networks (CNN). However, even the latest evolutions struggle to precisely delineating borders, which often leads to geometric distortions and inadvertent fusion of adjacent building instances. We propose to overcome this issue by exploiting the distinct geometric properties of buildings. To this end, we present Deep Structured Active Contours (DSAC), a novel framework that integrates priors and constraints into the segmentation process, such as continuous boundaries, smooth edges, and sharp corners. To do so, DSAC employs Active Contour Models (ACM), a family of constraint- and prior-based polygonal models. We learn ACM parameterizations per instance using a CNN, and show how to incorporate all components in a structured output model, making DSAC trainable end-to-end. We evaluate DSAC on three challenging building instance segmentation datasets, where it compares favorably against state-of-the-art. Code will be made available on https://github.com/dmarcosg/DSAC.

U2 - 10.1109/CVPR.2018.00925

DO - 10.1109/CVPR.2018.00925

M3 - Conference paper

SP - 8877

EP - 8885

BT - Proceedings - 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2018

PB - IEEE computer society

ER -

Marcos D, Tuia D, Kellenberger B, Zhang L, Bai M, Liao R et al. Learning Deep Structured Active Contours End-to-End. In Proceedings - 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2018. IEEE computer society. 2018. p. 8877-8885. 8579023 https://doi.org/10.1109/CVPR.2018.00925