Semi-Supervised Deep Learning Representations in Earth Observation Based Forest Management

Oleg Antropov; Matthieu Molinier; Ridvan Salih Kuzu; Lloyd Hughes; Marc Ruswurm; Devis Tuia; Corneliu Octavian Dumitru; Shaojia Ge; Sudipan Saha; Xiao Xiang Zhu

doi:10.1109/IGARSS52108.2023.10282283

Semi-Supervised Deep Learning Representations in Earth Observation Based Forest Management

Oleg Antropov^*
, Matthieu Molinier
, Ridvan Salih Kuzu
, Lloyd Hughes
, Marc Ruswurm
, Devis Tuia
, Corneliu Octavian Dumitru
, Shaojia Ge
, Sudipan Saha
, Xiao Xiang Zhu

^*Corresponding author for this work

BA6408 Remote sensing

German Aerospace Center (DLR)
Ecole Polytechnique Fédérale de Lausanne (EPFL)
Nanjing University of Science and Technology
Indian Institute of Technology, Delhi
Technical University of Munich (TUM)

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

4 Citations (Scopus)

Abstract

In this study, we examine the potential of several self-supervised deep learning models in predicting forest attributes and detecting forest changes using ESA Sentinel-1 and Sentinel-2 images. The performance of the proposed deep learning models is compared to established conventional machine learning approaches. Studied use-cases include mapping of forest disturbance (windthrown forests, snowload damages) using deep change vector analysis, forest height mapping using UNet+ based models, Momentum contrast and regression modeling. Study areas were represented by several boreal forest sites in Finland. Our results indicate that developed methods allow to achieve superior classification and prediction accuracies compared to traditional methodologies and mimimize the amount of necessary in-situ forestry data.

Original language	English
Title of host publication	IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings
Publisher	IEEE Institute of Electrical and Electronic Engineers
Pages	650-653
ISBN (Electronic)	979-8-3503-2010-7, 979-8-3503-2009-1
ISBN (Print)	979-8-3503-3174-5
DOIs	https://doi.org/10.1109/IGARSS52108.2023.10282283
Publication status	Published - 2023
MoE publication type	A4 Article in a conference publication
Event	2023 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023 - Pasadena, CA, USA, Pasadena, United States Duration: 16 Jul 2023 → 21 Jul 2023 https://2023.ieeeigarss.org/

Publication series

Series	IEEE International Geoscience and Remote Sensing Symposium Proceedings
ISSN	2153-6996

Conference

Conference	2023 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023
Country/Territory	United States
City	Pasadena
Period	16/07/23 → 21/07/23
Internet address	https://2023.ieeeigarss.org/

Funding

This work was financed by the European Space Agency under contract 4000137253/22/I-DT, Non-supervised representation learning for Sentinels (RepreSent).

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 12 Responsible Consumption and Production
SDG 15 Life on Land

Keywords

boreal zone
deep learning
forest height
forest management
satellite image
Sentinel-1
Sentinel-2

Access to Document

10.1109/IGARSS52108.2023.10282283

Cite this

Antropov, O., Molinier, M., Kuzu, R. S., Hughes, L., Ruswurm, M., Tuia, D., Dumitru, C. O., Ge, S., Saha, S., & Zhu, X. X. (2023). Semi-Supervised Deep Learning Representations in Earth Observation Based Forest Management. In IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings (pp. 650-653). IEEE Institute of Electrical and Electronic Engineers. https://doi.org/10.1109/IGARSS52108.2023.10282283

Antropov, Oleg ; Molinier, Matthieu ; Kuzu, Ridvan Salih et al. / Semi-Supervised Deep Learning Representations in Earth Observation Based Forest Management. IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings. IEEE Institute of Electrical and Electronic Engineers, 2023. pp. 650-653 (IEEE International Geoscience and Remote Sensing Symposium Proceedings).

@inproceedings{06e9e231a6024ce98d025a7059531553,

title = "Semi-Supervised Deep Learning Representations in Earth Observation Based Forest Management",

abstract = "In this study, we examine the potential of several self-supervised deep learning models in predicting forest attributes and detecting forest changes using ESA Sentinel-1 and Sentinel-2 images. The performance of the proposed deep learning models is compared to established conventional machine learning approaches. Studied use-cases include mapping of forest disturbance (windthrown forests, snowload damages) using deep change vector analysis, forest height mapping using UNet+ based models, Momentum contrast and regression modeling. Study areas were represented by several boreal forest sites in Finland. Our results indicate that developed methods allow to achieve superior classification and prediction accuracies compared to traditional methodologies and mimimize the amount of necessary in-situ forestry data.",

keywords = "boreal zone, deep learning, forest height, forest management, satellite image, Sentinel-1, Sentinel-2",

author = "Oleg Antropov and Matthieu Molinier and Kuzu, \{Ridvan Salih\} and Lloyd Hughes and Marc Ruswurm and Devis Tuia and Dumitru, \{Corneliu Octavian\} and Shaojia Ge and Sudipan Saha and Zhu, \{Xiao Xiang\}",

year = "2023",

doi = "10.1109/IGARSS52108.2023.10282283",

language = "English",

isbn = "979-8-3503-3174-5",

series = "IEEE International Geoscience and Remote Sensing Symposium Proceedings",

publisher = "IEEE Institute of Electrical and Electronic Engineers",

pages = "650--653",

booktitle = "IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings",

address = "United States",

note = "2023 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023 ; Conference date: 16-07-2023 Through 21-07-2023",

url = "https://2023.ieeeigarss.org/",

}

Antropov, O , Molinier, M, Kuzu, RS, Hughes, L, Ruswurm, M, Tuia, D, Dumitru, CO, Ge, S, Saha, S & Zhu, XX 2023, Semi-Supervised Deep Learning Representations in Earth Observation Based Forest Management. in IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings. IEEE Institute of Electrical and Electronic Engineers, IEEE International Geoscience and Remote Sensing Symposium Proceedings, pp. 650-653, 2023 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023, Pasadena, United States, 16/07/23. https://doi.org/10.1109/IGARSS52108.2023.10282283

Semi-Supervised Deep Learning Representations in Earth Observation Based Forest Management. / Antropov, Oleg ; Molinier, Matthieu; Kuzu, Ridvan Salih et al.
IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings. IEEE Institute of Electrical and Electronic Engineers, 2023. p. 650-653 (IEEE International Geoscience and Remote Sensing Symposium Proceedings).

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

TY - GEN

T1 - Semi-Supervised Deep Learning Representations in Earth Observation Based Forest Management

AU - Antropov, Oleg

AU - Molinier, Matthieu

AU - Kuzu, Ridvan Salih

AU - Hughes, Lloyd

AU - Ruswurm, Marc

AU - Tuia, Devis

AU - Dumitru, Corneliu Octavian

AU - Ge, Shaojia

AU - Saha, Sudipan

AU - Zhu, Xiao Xiang

PY - 2023

Y1 - 2023

N2 - In this study, we examine the potential of several self-supervised deep learning models in predicting forest attributes and detecting forest changes using ESA Sentinel-1 and Sentinel-2 images. The performance of the proposed deep learning models is compared to established conventional machine learning approaches. Studied use-cases include mapping of forest disturbance (windthrown forests, snowload damages) using deep change vector analysis, forest height mapping using UNet+ based models, Momentum contrast and regression modeling. Study areas were represented by several boreal forest sites in Finland. Our results indicate that developed methods allow to achieve superior classification and prediction accuracies compared to traditional methodologies and mimimize the amount of necessary in-situ forestry data.

AB - In this study, we examine the potential of several self-supervised deep learning models in predicting forest attributes and detecting forest changes using ESA Sentinel-1 and Sentinel-2 images. The performance of the proposed deep learning models is compared to established conventional machine learning approaches. Studied use-cases include mapping of forest disturbance (windthrown forests, snowload damages) using deep change vector analysis, forest height mapping using UNet+ based models, Momentum contrast and regression modeling. Study areas were represented by several boreal forest sites in Finland. Our results indicate that developed methods allow to achieve superior classification and prediction accuracies compared to traditional methodologies and mimimize the amount of necessary in-situ forestry data.

KW - boreal zone

KW - deep learning

KW - forest height

KW - forest management

KW - satellite image

KW - Sentinel-1

KW - Sentinel-2

UR - https://www.scopus.com/pages/publications/85178341350

U2 - 10.1109/IGARSS52108.2023.10282283

DO - 10.1109/IGARSS52108.2023.10282283

M3 - Conference article in proceedings

AN - SCOPUS:85178341350

SN - 979-8-3503-3174-5

T3 - IEEE International Geoscience and Remote Sensing Symposium Proceedings

SP - 650

EP - 653

BT - IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings

PB - IEEE Institute of Electrical and Electronic Engineers

T2 - 2023 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023

Y2 - 16 July 2023 through 21 July 2023

ER -

Antropov O , Molinier M, Kuzu RS, Hughes L, Ruswurm M, Tuia D et al. Semi-Supervised Deep Learning Representations in Earth Observation Based Forest Management. In IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings. IEEE Institute of Electrical and Electronic Engineers. 2023. p. 650-653. (IEEE International Geoscience and Remote Sensing Symposium Proceedings). doi: 10.1109/IGARSS52108.2023.10282283

Semi-Supervised Deep Learning Representations in Earth Observation Based Forest Management

Abstract

Publication series

Conference

Funding

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this