1. APLICAÇÃO DE APRENDIZADO DE MÁQUINA COM DADOS DE SENSORIAMENTO REMOTO PARA O MAPEAMENTO DE FLORESTAS URBANAS.
- Author
-
Guimarães Cano, Priscila Lôpo and Marcato Junior, José
- Subjects
- *
RANDOM forest algorithms , *SUSTAINABLE urban development , *FOREST conservation , *ENVIRONMENTAL quality , *ENVIRONMENTAL indicators , *URBAN forestry , *URBAN health , *IMAGE segmentation - Abstract
Urban forests provide several benefits to cities, including lowering temperatures, improving air quality, health and leisure for the population, and protecting watersheds, thus making them one of the most important indicators of environmental quality and urban sustainability. Campo Grande, in Mato Grosso do Sul, has the title of "Tree Cities of the World", which recognizes the cities most committed to the preservation of urban forests and sustainable development, therefore mapping and monitoring serve as an aid tool for governments and decision makers. The present work consisted of combining high resolution remote sensing images and machine learning algorithms to map urban forests. The study was carried out in the Prosa Watershed, Campo Grande, Mato Grosso do Sul, Brazil, considering Google Earth images from May 14, 2020. For classification purposes, the Random Forest algorithm associated with previous image segmentation was adopted. with the mean shift technique. As a result, a percentage of 18.31% of arboreal vegetation in the watershed was obtained and the F1 metric was higher than 85%, thus enabling an accurate and updated mapping of urban forests. [ABSTRACT FROM AUTHOR]
- Published
- 2021
- Full Text
- View/download PDF