Deforestation via Unsupervised Computing Learning: Modeling and Applications in Preservation Parks of the Amazon Biome

• Grant number: 22/13665-0
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: January 01, 2023
• End date: December 31, 2023
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Wallace Correa de Oliveira Casaca
• Grantee: Júlia Rodrigues Marques do Nascimento
• Host Institution: Instituto de Biociências, Letras e Ciências Exatas (IBILCE). Universidade Estadual Paulista (UNESP). Campus de São José do Rio Preto. São José do Rio Preto , SP, Brazil

Abstract

Computational Learning-inspired techniques has contributed effectively and transdiciplinarily in Computational Vision and Pattern Recognition applications, such as deforstation mapping in the Amazon biome. In this context, data-driven algorithms have promoted the development of innovative and accurate tools for the detection of deforested areas, by analyzing time series of remotely sensed images while gaining knowledge about the current state of the target region and its historical vegetation cover data. Considering the above-discussed issue, this research focuses on the study and implementation of a Computational Learning-based methodology to classify deforested areas in the Brazilian Amazon. Our goal is to study, assess and investigate time-series of deforestation data concerning environmental preservation units in São Félix do Xingu, Brazil's sixth largest municipality in Brazil, located in the state of Pará. The proposed data-driven solution combine the use of fully unsupervised Machine Learning techniques for dealing with image time-series, i.e., the One-Class Support Vector Machine and Isolation Forest methods, as well as Anomaly Detection models and Spectral Indices so as to properly identify and classify deforested areas. Besides the robustness and accuracy provided when applying Machine Learning, the proposed methodology has been designed to work as a user-independent tool in aspects such as the non-mandatory use of reference data and the capability to access integrated, up-to-date catalogs from the Google Earth Engine open access platform.

Scientific publications

(The scientific publications listed on this page originate from the Web of Science or SciELO databases. Their authors have cited FAPESP grant or fellowship project numbers awarded to Principal Investigators or Fellowship Recipients, whether or not they are among the authors. This information is collected automatically and retrieved directly from those bibliometric databases.)

RODRIGUES, JULIA; DIAS, MAURICIO ARAUJO; NEGRI, ROGERIO; HUSSAIN, SARDAR MUHAMMAD; CASACA, WALLACE. A Robust Dual-Mode Machine Learning Framework for Classifying Deforestation Patterns in Amazon Native Lands. LAND, v. 13, n. 9, p. 19-pg., 2024-09-01. (22/13665-0, 13/07375-0, 23/14427-8)