Automatic segmentation of river and land in SAR images: A deep learning approach

M. M. Manohara Pai, Vaibhav Mehrotra, Shreyas Aiyar, Ujjwal Verma, Radhika M. Pai

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

The ubiquitousness of satellite imagery and powerful, computationally efficient Deep Learning frameworks have found profound use in the field of remote sensing. Augmented with easy access to abundant image data made available by different satellites such as LANDSAT and European Space Agency's Copernicus missions, deep learning has opened various avenues of research in monitoring the world's oceans, land, rivers, etc. One significant problem in this direction is the accurate identification and subsequent segmentation of surface-water in images in the microwave spectrum. Typically, standard image processing tools are used to segment the images which are time inefficient. However, in recent years, deep learning methods for semantic segmentation is the preferred choice given its high accuracy and ease of use. This paper proposes the use of deep-learning approaches such as U-Net to perform an efficient segmentation of river and land. Experimental results show that our approach achieves vastly superior performance on SAR images with pixel accuracy of 0.98 and F1 score of 0.99.

Original languageEnglish
Title of host publicationProceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages15-20
Number of pages6
ISBN (Electronic)9781728114880
DOIs
Publication statusPublished - 06-2019
Event2nd IEEE International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019 - Cagliari, Sardinia, Italy
Duration: 03-06-201905-06-2019

Publication series

NameProceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019

Conference

Conference2nd IEEE International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019
CountryItaly
CityCagliari, Sardinia
Period03-06-1905-06-19

Fingerprint

Rivers
Satellite imagery
Surface waters
Remote sensing
Image processing
Pixels
Semantics
Microwaves
Satellites
Deep learning
Segmentation
Monitoring

All Science Journal Classification (ASJC) codes

  • Artificial Intelligence
  • Computer Science Applications
  • Information Systems and Management

Cite this

Manohara Pai, M. M., Mehrotra, V., Aiyar, S., Verma, U., & Pai, R. M. (2019). Automatic segmentation of river and land in SAR images: A deep learning approach. In Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019 (pp. 15-20). [8791719] (Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/AIKE.2019.00011
Manohara Pai, M. M. ; Mehrotra, Vaibhav ; Aiyar, Shreyas ; Verma, Ujjwal ; Pai, Radhika M. / Automatic segmentation of river and land in SAR images : A deep learning approach. Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 15-20 (Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019).
@inproceedings{6cb228f3e7944f7490732bd7b3c3ed23,
title = "Automatic segmentation of river and land in SAR images: A deep learning approach",
abstract = "The ubiquitousness of satellite imagery and powerful, computationally efficient Deep Learning frameworks have found profound use in the field of remote sensing. Augmented with easy access to abundant image data made available by different satellites such as LANDSAT and European Space Agency's Copernicus missions, deep learning has opened various avenues of research in monitoring the world's oceans, land, rivers, etc. One significant problem in this direction is the accurate identification and subsequent segmentation of surface-water in images in the microwave spectrum. Typically, standard image processing tools are used to segment the images which are time inefficient. However, in recent years, deep learning methods for semantic segmentation is the preferred choice given its high accuracy and ease of use. This paper proposes the use of deep-learning approaches such as U-Net to perform an efficient segmentation of river and land. Experimental results show that our approach achieves vastly superior performance on SAR images with pixel accuracy of 0.98 and F1 score of 0.99.",
author = "{Manohara Pai}, {M. M.} and Vaibhav Mehrotra and Shreyas Aiyar and Ujjwal Verma and Pai, {Radhika M.}",
year = "2019",
month = "6",
doi = "10.1109/AIKE.2019.00011",
language = "English",
series = "Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "15--20",
booktitle = "Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019",
address = "United States",

}

Manohara Pai, MM, Mehrotra, V, Aiyar, S, Verma, U & Pai, RM 2019, Automatic segmentation of river and land in SAR images: A deep learning approach. in Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019., 8791719, Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019, Institute of Electrical and Electronics Engineers Inc., pp. 15-20, 2nd IEEE International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019, Cagliari, Sardinia, Italy, 03-06-19. https://doi.org/10.1109/AIKE.2019.00011

Automatic segmentation of river and land in SAR images : A deep learning approach. / Manohara Pai, M. M.; Mehrotra, Vaibhav; Aiyar, Shreyas; Verma, Ujjwal; Pai, Radhika M.

Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019. Institute of Electrical and Electronics Engineers Inc., 2019. p. 15-20 8791719 (Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Automatic segmentation of river and land in SAR images

T2 - A deep learning approach

AU - Manohara Pai, M. M.

AU - Mehrotra, Vaibhav

AU - Aiyar, Shreyas

AU - Verma, Ujjwal

AU - Pai, Radhika M.

PY - 2019/6

Y1 - 2019/6

N2 - The ubiquitousness of satellite imagery and powerful, computationally efficient Deep Learning frameworks have found profound use in the field of remote sensing. Augmented with easy access to abundant image data made available by different satellites such as LANDSAT and European Space Agency's Copernicus missions, deep learning has opened various avenues of research in monitoring the world's oceans, land, rivers, etc. One significant problem in this direction is the accurate identification and subsequent segmentation of surface-water in images in the microwave spectrum. Typically, standard image processing tools are used to segment the images which are time inefficient. However, in recent years, deep learning methods for semantic segmentation is the preferred choice given its high accuracy and ease of use. This paper proposes the use of deep-learning approaches such as U-Net to perform an efficient segmentation of river and land. Experimental results show that our approach achieves vastly superior performance on SAR images with pixel accuracy of 0.98 and F1 score of 0.99.

AB - The ubiquitousness of satellite imagery and powerful, computationally efficient Deep Learning frameworks have found profound use in the field of remote sensing. Augmented with easy access to abundant image data made available by different satellites such as LANDSAT and European Space Agency's Copernicus missions, deep learning has opened various avenues of research in monitoring the world's oceans, land, rivers, etc. One significant problem in this direction is the accurate identification and subsequent segmentation of surface-water in images in the microwave spectrum. Typically, standard image processing tools are used to segment the images which are time inefficient. However, in recent years, deep learning methods for semantic segmentation is the preferred choice given its high accuracy and ease of use. This paper proposes the use of deep-learning approaches such as U-Net to perform an efficient segmentation of river and land. Experimental results show that our approach achieves vastly superior performance on SAR images with pixel accuracy of 0.98 and F1 score of 0.99.

UR - http://www.scopus.com/inward/record.url?scp=85071443420&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85071443420&partnerID=8YFLogxK

U2 - 10.1109/AIKE.2019.00011

DO - 10.1109/AIKE.2019.00011

M3 - Conference contribution

AN - SCOPUS:85071443420

T3 - Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019

SP - 15

EP - 20

BT - Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Manohara Pai MM, Mehrotra V, Aiyar S, Verma U, Pai RM. Automatic segmentation of river and land in SAR images: A deep learning approach. In Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019. Institute of Electrical and Electronics Engineers Inc. 2019. p. 15-20. 8791719. (Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019). https://doi.org/10.1109/AIKE.2019.00011