Identification of crop-livestock integration systems (CLI) in time series of high resolution imagery

• Grant number: 19/16870-0
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: December 01, 2019
• End date: November 30, 2020
• Field of knowledge: Agronomical Sciences - Agricultural Engineering
• Principal Investigator: Aliny Aparecida dos Reis
• Grantee: Henrique Sandes Lima Almeida
• Host Institution: Núcleo Interdisciplinar de Planejamento Energético (NIPE). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil

Abstract

Increasing agricultural production to meet world food demand in a sustainable manner in the available agricultural lands is one of the biggest challenges for crops and livestock activities in the present time. In this sense, the crop-livestock integration (CLI) systems stand out for allowing sustainable production from the integration of agricultural and livestock activities in the same area. Brazil is a pioneer country in CLI systems, however, given the difficulties in identifying these areas, there is no statistic of their proportion in relation to traditional agriculture and livestock. Time series of remote sensing imagery is one of the most efficient ways of identifying and monitoring these systems. Several methods have been developed for identifying targets in time series of remote sensing imagery; however, deep learning methods represent the state-of-the-art in a variety of domains, including the problems of identifying targets in time series. Since crop-livestock integration systems are highly dynamic cropping systems, mapping and monitoring these systems using remote sensing imagery is still a challenge, even when robust methods are used. There are no reports of the application of deep learning algorithms to identify integration systems in time series of high spatial and temporal remote sensing imagery. Given this scenario, this project aims to study deep learning methods for the identification of crop-livestock integration systems in the time series of high-resolution images. The results of this study are expected to provide a methodology for the classification of CLI systems in high-resolution remote sensing imagery using deep learning methods. (AU)