Detection of orange plantation lines using high-resolution orthomosaic and convolutional neural network

• Grant number: 23/02772-2
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: April 01, 2023
• End date: March 31, 2024
• Field of knowledge: Physical Sciences and Mathematics - Geosciences - Geodesy
• Principal Investigator: Aluir Porfírio Dal Poz
• Grantee: Letícia Rodrigues dos Santos
• Host Institution: Faculdade de Ciências e Tecnologia (FCT). Universidade Estadual Paulista (UNESP). Campus de Presidente Prudente. Presidente Prudente , SP, Brazil
• Associated research grant: 21/06029-7 - High resolution remote sensing for digital agriculture, AP.TEM

Abstract

This working plan proposes the use of orthomosaics, produced from very high resolution (circa 10 cm) RGB images, acquired via the Unmanned Aerial Vehicle (UAV) platform, to detect rows or lines of orange plantation. The image analysis method to be used is based on deep learning, known as convolutional neural networks (CNN). The basic idea to be explored in CNN learning is that there will be no plants outside the plantation line and, consequently, plantation lines cannot be formed without there being plants. The orthomosaic to be used in the experiments corresponds to a rural property located in the municipality of Santa Cruz do Rio Pardo-SP. This orthomosaic shows a field of orange plantation lines, where the plants usually appear connected to each other. The results obtained will be evaluated based on quantitative quality metrics, such as the IoU (Intersection over Union), Precision, Recall, and F1-score.