Banca de QUALIFICAÇÃO: JOYCE DE AGUIAR CARVALHO

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : JOYCE DE AGUIAR CARVALHO
DATE: 09/12/2020
TIME: 08:30
LOCAL: Instituto de Tecnologia/ Departamento de Engenharia
TITLE:

CHARACTERIZATION OF WEED PLANTS IN THE CULTURE OF SORGHUM (SORGHUM BICOLOR) BY PORTABLE SPECTRORADIOMETER AND REMOTELY PILOTED AIRCRAFT (RPAS)

KEY WORDS:

Spectral signature, Precision agriculture, Multivariate analysis.

PAGES: 17
BIG AREA: Ciências Agrárias
AREA: Engenharia Agrícola
SUBÁREA: Máquinas e Implementos Agrícolas
SUMMARY:

Sorghum is a rustic plant that stands out for its high biomass production and tolerance to water deficit making it a good alternative for agricultural diversification. A latent factor that causes problems in the production of sorghum is the presence of weeds. The sooner weed populations are detected and suppressed from the area, the smaller the damage to the installed culture. The use of Remote Sensing techniques by spectroradiometer and by means of images made by Remote Piloted Aircraft (RPAs) has become a great tool in Precision Agriculture, making decision-making for crop management faster and more accurate. The present work aims to define the appropriate spectral range to differentiate sorghum weeds by means of the spectroradiometer and to use RGB images extracted from RPAs to detect weeds by means of image classification techniques for sorghum culture. The experiment will be carried out at UFRRJ, located in Seropédica RJ, in an area of 9 ha of sorghum production. Spectral characterization will be performed for weeds and for sorghum culture by an ASD FIELDSPEC® 4 spectroradiometer. Spectral readings will be analyzed according to the component analysis method (PCA), in order to obtain consistent classifications regarding the discrimination of spectral bands. For the flight an ARP model Dji Phanton 4 Pro will be used, shipped with an RGB camera. After being processed, the images will go through supervised and unsupervised classifications, by the Maximum Likelihood and Isoseg classifiers, respectively. To compare the accuracy of the maps between the classifiers, the global accuracy indices and Kappa coefficient will be used. It is expected in the present work to detect the appropriate spectral range to differentiate weeds from sorghum and produce maps with speed and accuracy for weed detection, allowing greater efficiency in the productive sector.

BANKING MEMBERS:
Presidente - 1050615 - MURILO MACHADO DE BARROS
Interno - 2161955 - ANDERSON GOMIDE COSTA
Externo à Instituição - FLÁVIO CASTRO DA SILVA - UFF
Externo à Instituição - GABRIEL ARAUJO E SILVA FERRAZ - UFLA