A Satellite Band Selection Framework for Amazon Forest Deforestation Detection Task

arXiv:2404.02659v1 Announce Type: new
Abstract: The conservation of tropical forests is a topic of significant social and ecological relevance due to their crucial role in the global ecosystem. Unfortunately, deforestation and degradation impact millions of hectares annually, necessitating government or private initiatives for effective forest monitoring. This study introduces a novel framework that employs the Univariate Marginal Distribution Algorithm (UMDA) to select spectral bands from Landsat-8 satellite, optimizing the representation of deforested areas. This selection guides a semantic segmentation architecture, …

abstract amazon arxiv conservation cs.cv cs.ne deforestation detection ecosystem forests framework global government impact monitoring role satellite social study type