Deforestation Detection With Weak Supervised Convolutional Neural Networks In Tropical Biomes

Type Article
Date 2022
Language English
Author(s) Soto Vega Pedro Juan1, Costa G. A. O. P.2, Ortega M. X.3, Bermudez J. D.3, Feitosa R. Q.3
Affiliation(s) 1 : Ifremer, PDG-REM-EEP-LEP, F-29280 Plouzan´e, France
2 : Dept. of Informatics and Computer Science, Rio de Janeiro State University (UERJ), Brazil
3 : Dept. of Electrical Engineering, Pontifical Catholic University of Rio de Janeiro (PUC-Rio), Brazil
Meeting XXIV ISPRS Congress (2022 edition) : Imaging Today, Foreseeing Tomorrow, Commission III, 6–11 June 2022, Nice, France
Source The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences (1682-1750) (Copernicus GmbH), 2022 , Vol. XLIII-B3-2022 , P. 713-719
DOI 10.5194/isprs-archives-XLIII-B3-2022-713-2022
Keyword(s) Change detection, deep learning, domain adaptation, deforestation, weak supervision

Deep learning methods are known to demand large amounts of labeled samples for training. For remote sensing applications such as change detection, coping with that demand is expensive and time-consuming. This work aims at investigating a noisy-label-based weak supervised method in the context of a deforestation mapping application, characterized by a high class imbalance between the classes of interest, i.e., deforestation and no-deforestation. The study sites correspond to different regions in the Amazon and Brazilian Cerrado biomes. To mitigate the lack of ground-truth labeled training samples, we devised an unsupervised pseudo-labeling scheme based on the Change Vector Analysis technique. The experimental results indicate that the proposed approach can improve the accuracy of deforestation detection applications.

Full Text
File Pages Size Access
Publisher's official version 7 2 MB Open access
Top of the page