Domain-Adversarial Neural Networks for Deforestation Detection in Tropical Forests

Many deep-learning-based, domain adaptation methods for remote sensing applications rely on adversarial training strategies to align features extracted from images of different domains in a shared latent space. However, the performance of such representation matching techniques is negatively impacte...

Full description

Saved in:

Bibliographic Details
Published in:	IEEE geoscience and remote sensing letters Vol. 19; p. 1
Main Authors:	Soto, Pedro J., Costa, Gilson A., Feitosa, Raul Q., Ortega, Mabel X., Bermudez, Jose D., Turnes, Javier N.
Format:	Journal Article
Language:	English
Published:	Piscataway IEEE 01-01-2022 The Institute of Electrical and Electronics Engineers, Inc. (IEEE) IEEE - Institute of Electrical and Electronics Engineers
Subjects:	Change Detection Computer Science Deep Learning Deforestation Deforestation Detection Detection Domain Adaptation Domains Environmental Engineering Environmental Sciences Feature extraction Forestry Image Processing Matching Neural and Evolutionary Computing Neural networks Neurons Noise measurement Remote Sensing Representations Task analysis Training Tropical climate Tropical forests Vector analysis Deforestation Detection Remote Sensing Domain Adaptation Deep Learning Change Detection
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	Many deep-learning-based, domain adaptation methods for remote sensing applications rely on adversarial training strategies to align features extracted from images of different domains in a shared latent space. However, the performance of such representation matching techniques is negatively impacted when class occurrences in the target domain, for which no labelled data is available during training, are highly imbalanced. In this work, we propose a deep-learning-based representation matching approach for domain adaptation in the context of change detection tasks. We further evaluate the approach in a deforestation mapping application, characterized by a high-class imbalance between the deforestation and no-deforestation classes. The domains represent different sites in the Amazon and Brazilian Cerrado biomes. To mitigate the class imbalance problem, we devised an unsupervised pseudo-labeling scheme based on Change Vector Analysis that prevents the feature alignment to be biased towards the over-represented class. The experimental results indicate that the proposed approach can improve the accuracy of cross-domain deforestation detection.
ISSN:	1545-598X 1558-0571
DOI:	10.1109/LGRS.2022.3163575