Bibliographic citations
Tovar, J., (2024). Estimación del carbono almacenado en paisajes agropecuarios y ganaderos a partir de Imágenes multiespectrales capturadas por drones, y técnicas de aprendizaje profundo [Pontificia Universidad Católica del Perú]. http://hdl.handle.net/20.500.12404/26749
Tovar, J., Estimación del carbono almacenado en paisajes agropecuarios y ganaderos a partir de Imágenes multiespectrales capturadas por drones, y técnicas de aprendizaje profundo []. PE: Pontificia Universidad Católica del Perú; 2024. http://hdl.handle.net/20.500.12404/26749
@mastersthesis{renati/529850,
title = "Estimación del carbono almacenado en paisajes agropecuarios y ganaderos a partir de Imágenes multiespectrales capturadas por drones, y técnicas de aprendizaje profundo",
author = "Tovar Galarreta, Juan Carlos Alfredo",
publisher = "Pontificia Universidad Católica del Perú",
year = "2024"
}
Agricultural and livestock landscapes are closely related to carbon flux, serving as carbon 1 reservoirs in the form of biomass.The assessment of carbon stocks stored in these areas is important to 2 support decision-making that prevents these areas from becoming carbone misión sources. However, 3 traditional carbon stock estimation techniques require on-site sampling, which is a strenuous, risky, 4 expensive and low-coverage task. In this study, we propose a methodology, based on multispectral 5 images captured by drones and Deep learning models, to automate the task of estimating maps of 6 carbon stocks sequestered in agricultural and livestock landscapes. We applied U-Net neural network 7 architecture to discriminate arboreal and grasslands zones. Then, a convolutional neural-network 8 based model is developed for carbon density estimation from multispectral images. Experimental 9 results on agricultural and livestock landscapes in Peruvian Amazon regions showed the effectiveness 10 of the proposed methodology, reporting mIoU of 87%, RMSE of 2.44 for arboreal zones, and RMSE 11 of 1.84 for grassland zones. We conclude that the carbon density estimation is achievable with the 12 proposed approach. This methodology can be helpful for decision-making and may contribute to the 13 management or controlling climate change.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.