Toward country scale building detection with convolutional neural network using aerial images

Hsiuhan Lexie Yang; Dalton Lunga; Jiangye Yuan

doi:10.1109/IGARSS.2017.8127091

Toward country scale building detection with convolutional neural network using aerial images

Hsiuhan Lexie Yang, Dalton Lunga, Jiangye Yuan

GeoAI

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

18 Scopus citations

Abstract

Establishing up-to-date nationwide building maps is essential to understand urban dynamics, such as estimating population and urban planning and many other applications. However, an efficient and effective solution is yet to be developed. In this paper, for the first time we evaluate three state-of-the-art CNNs for detecting buildings across entire United States using aerial images. The three CNN architectures, fully convolutional neural network, conditional random field as recurrent neural network, and SegNet, support semantic pixel-wise labeling and focus on capturing textural information at multi-scale. We use 1-meter resolution NAIP images as the test data set, and compare the detection results across the three methods. In addition, we propose to combine signed distance function labels with SegNet, which is the preferred CNN architecture identified by our extensive evaluations. The results are further improved in terms of precision, recall rate and the number of building detected. On average, model inference on test images is less than one minute for an area of size ∼ 56 km². With these promising results and the time required to process images, the framework offers great potential toward country scale building mapping with remote sensing imagery.

Original language	English
Title of host publication	2017 IEEE International Geoscience and Remote Sensing Symposium
Subtitle of host publication	International Cooperation for Global Awareness, IGARSS 2017 - Proceedings
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	870-873
Number of pages	4
ISBN (Electronic)	9781509049516
DOIs	https://doi.org/10.1109/IGARSS.2017.8127091
State	Published - Dec 1 2017
Event	37th Annual IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2017 - Fort Worth, United States Duration: Jul 23 2017 → Jul 28 2017

Publication series

Name	International Geoscience and Remote Sensing Symposium (IGARSS)
Volume	2017-July

Conference

Conference	37th Annual IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2017
Country/Territory	United States
City	Fort Worth
Period	07/23/17 → 07/28/17

Funding

This manuscript has been authored by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725 with the U.S. De- partment of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a nonexclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. This manuscript has been authored by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725 with the U.S. Department of Energy.

Keywords

NAIP
building extractions
convolutional
deep learning
semantic segmentation

Access to Document

10.1109/IGARSS.2017.8127091

Cite this

Yang, H. L., Lunga, D., & Yuan, J. (2017). Toward country scale building detection with convolutional neural network using aerial images. In 2017 IEEE International Geoscience and Remote Sensing Symposium: International Cooperation for Global Awareness, IGARSS 2017 - Proceedings (pp. 870-873). Article 8127091 (International Geoscience and Remote Sensing Symposium (IGARSS); Vol. 2017-July). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IGARSS.2017.8127091

Yang, Hsiuhan Lexie ; Lunga, Dalton ; Yuan, Jiangye. / Toward country scale building detection with convolutional neural network using aerial images. 2017 IEEE International Geoscience and Remote Sensing Symposium: International Cooperation for Global Awareness, IGARSS 2017 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2017. pp. 870-873 (International Geoscience and Remote Sensing Symposium (IGARSS)).

@inproceedings{1aaf1e0650c14560bd6b7107bf173a7f,

title = "Toward country scale building detection with convolutional neural network using aerial images",

abstract = "Establishing up-to-date nationwide building maps is essential to understand urban dynamics, such as estimating population and urban planning and many other applications. However, an efficient and effective solution is yet to be developed. In this paper, for the first time we evaluate three state-of-the-art CNNs for detecting buildings across entire United States using aerial images. The three CNN architectures, fully convolutional neural network, conditional random field as recurrent neural network, and SegNet, support semantic pixel-wise labeling and focus on capturing textural information at multi-scale. We use 1-meter resolution NAIP images as the test data set, and compare the detection results across the three methods. In addition, we propose to combine signed distance function labels with SegNet, which is the preferred CNN architecture identified by our extensive evaluations. The results are further improved in terms of precision, recall rate and the number of building detected. On average, model inference on test images is less than one minute for an area of size ∼ 56 km2. With these promising results and the time required to process images, the framework offers great potential toward country scale building mapping with remote sensing imagery.",

keywords = "NAIP, building extractions, convolutional, deep learning, semantic segmentation",

author = "Yang, {Hsiuhan Lexie} and Dalton Lunga and Jiangye Yuan",

note = "Publisher Copyright: {\textcopyright} 2017 IEEE.; 37th Annual IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2017 ; Conference date: 23-07-2017 Through 28-07-2017",

year = "2017",

month = dec,

day = "1",

doi = "10.1109/IGARSS.2017.8127091",

language = "English",

series = "International Geoscience and Remote Sensing Symposium (IGARSS)",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "870--873",

booktitle = "2017 IEEE International Geoscience and Remote Sensing Symposium",

}

Yang, HL , Lunga, D & Yuan, J 2017, Toward country scale building detection with convolutional neural network using aerial images. in 2017 IEEE International Geoscience and Remote Sensing Symposium: International Cooperation for Global Awareness, IGARSS 2017 - Proceedings., 8127091, International Geoscience and Remote Sensing Symposium (IGARSS), vol. 2017-July, Institute of Electrical and Electronics Engineers Inc., pp. 870-873, 37th Annual IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2017, Fort Worth, United States, 07/23/17. https://doi.org/10.1109/IGARSS.2017.8127091

Toward country scale building detection with convolutional neural network using aerial images. / Yang, Hsiuhan Lexie ; Lunga, Dalton; Yuan, Jiangye.
2017 IEEE International Geoscience and Remote Sensing Symposium: International Cooperation for Global Awareness, IGARSS 2017 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2017. p. 870-873 8127091 (International Geoscience and Remote Sensing Symposium (IGARSS); Vol. 2017-July).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Toward country scale building detection with convolutional neural network using aerial images

AU - Yang, Hsiuhan Lexie

AU - Lunga, Dalton

AU - Yuan, Jiangye

PY - 2017/12/1

Y1 - 2017/12/1

N2 - Establishing up-to-date nationwide building maps is essential to understand urban dynamics, such as estimating population and urban planning and many other applications. However, an efficient and effective solution is yet to be developed. In this paper, for the first time we evaluate three state-of-the-art CNNs for detecting buildings across entire United States using aerial images. The three CNN architectures, fully convolutional neural network, conditional random field as recurrent neural network, and SegNet, support semantic pixel-wise labeling and focus on capturing textural information at multi-scale. We use 1-meter resolution NAIP images as the test data set, and compare the detection results across the three methods. In addition, we propose to combine signed distance function labels with SegNet, which is the preferred CNN architecture identified by our extensive evaluations. The results are further improved in terms of precision, recall rate and the number of building detected. On average, model inference on test images is less than one minute for an area of size ∼ 56 km2. With these promising results and the time required to process images, the framework offers great potential toward country scale building mapping with remote sensing imagery.

AB - Establishing up-to-date nationwide building maps is essential to understand urban dynamics, such as estimating population and urban planning and many other applications. However, an efficient and effective solution is yet to be developed. In this paper, for the first time we evaluate three state-of-the-art CNNs for detecting buildings across entire United States using aerial images. The three CNN architectures, fully convolutional neural network, conditional random field as recurrent neural network, and SegNet, support semantic pixel-wise labeling and focus on capturing textural information at multi-scale. We use 1-meter resolution NAIP images as the test data set, and compare the detection results across the three methods. In addition, we propose to combine signed distance function labels with SegNet, which is the preferred CNN architecture identified by our extensive evaluations. The results are further improved in terms of precision, recall rate and the number of building detected. On average, model inference on test images is less than one minute for an area of size ∼ 56 km2. With these promising results and the time required to process images, the framework offers great potential toward country scale building mapping with remote sensing imagery.

KW - NAIP

KW - building extractions

KW - convolutional

KW - deep learning

KW - semantic segmentation

UR - http://www.scopus.com/inward/record.url?scp=85034400093&partnerID=8YFLogxK

U2 - 10.1109/IGARSS.2017.8127091

DO - 10.1109/IGARSS.2017.8127091

M3 - Conference contribution

AN - SCOPUS:85034400093

T3 - International Geoscience and Remote Sensing Symposium (IGARSS)

SP - 870

EP - 873

BT - 2017 IEEE International Geoscience and Remote Sensing Symposium

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 37th Annual IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2017

Y2 - 23 July 2017 through 28 July 2017

ER -

Yang HL , Lunga D, Yuan J. Toward country scale building detection with convolutional neural network using aerial images. In 2017 IEEE International Geoscience and Remote Sensing Symposium: International Cooperation for Global Awareness, IGARSS 2017 - Proceedings. Institute of Electrical and Electronics Engineers Inc. 2017. p. 870-873. 8127091. (International Geoscience and Remote Sensing Symposium (IGARSS)). doi: 10.1109/IGARSS.2017.8127091

Toward country scale building detection with convolutional neural network using aerial images

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this