Using deep learning to identify invasive Pinus spp. in wetlands

• Grant number: 22/11438-6
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: March 01, 2023
• End date: February 29, 2024
• Field of knowledge: Agronomical Sciences - Forestry Resources and Forestry Engineering - Nature Conservation
• Principal Investigator: Paulo Guilherme Molin
• Grantee: Giovanna de Andrade Ferreira
• Host Institution: Centro de Ciências da Natureza (CCN). Universidade Federal de São Carlos (UFSCAR). Campus de Lagoa do Sino. Buri , SP, Brazil
• Associated research grant: 18/18416-2 - Understanding restored forests for benefiting people and nature - NewFor, AP.BTA.TEM

Abstract

Invasive exotic species are plants, animals, or exotic organisms introduced into native ecosystems by breaking down geographic barriers, whether anthropogenic or natural. These exotic species can interfere negatively in the natural functioning of the ecosystem to which they were inserted and modify the natural ecological relationships. These species can acquire advantages in the dispute for food and habitat, providing their survival and rapid reproduction or dispersion. The introduction of Pinus spp. in Brazil is an example of such invasive potential, which caused a series of environmental disorders and led to the impoverishment of native vegetation biodiversity in some regions. When an exotic species with invasive potential is detected, immediate actions must be taken to avoid negative (and sometimes irreversible) impacts on biodiversity and ecosystem services. In order to assess and identify the invasive species, the demand for tools capable of providing the location and quantification of individuals becomes urgent. Therefore, the goal of this work is to use Mask R-CNN algorithms (Regional Convolutional Neural Mask Network) as a deep learning tool (through the segmentation UAV high resolution images) to identify Pinus spp. individuals in native wetland area, located on the Lagoa do Sino campus of the Federal University of São Carlos. The results obtained with this study will contribute to optimize the diagnosis of biological invasion, allowing to quantify the number of individuals and their spatial distribution, collaborating for the subsequent methodology of more effective management plans for the eradication of invasive individuals and therefore allowing the natural regeneration of the ecosystem.