Improvement of the algorithm and digital tool for crop estimation in arabica coffee crops and its implementation in a digital platform

• Grant number: 23/09595-9
• Support Opportunities: Research Grants - Innovative Research in Small Business - PIPE
• Start date: February 01, 2024
• End date: January 31, 2026
• Field of knowledge: Agronomical Sciences - Agronomy - Crop Science
• Principal Investigator: Guilherme Jorge Gomes de Sousa
• Grantee: Guilherme Jorge Gomes de Sousa
• Company: Agrobee Soluções em Polinização e Sustentabilidade Ltda
• CNAE: Pesquisa e desenvolvimento experimental em ciências físicas e naturais
• Pesquisadores principais: Diego Moure Oliveira ; Joyce Mayra Volpini Almeida Dias
• Associated researchers: Andresa Aparecida Berretta e Silva ; Carlos Pamplona Rehder ; Jorge Luis Silva Kavicki
• Associated research grant: 22/05594-5 - Digital platform for forecasting and harvest in arabica coffee crops, AP.PIPE
• Associated scholarship(s): 25/10501-4 - Development of a digital tool for crop estimation in Arabica coffee plantations, BP.TT; 24/09227-2 - Analysis of environmental data to correlate with harvest estimates in coffee plantations, BP.TT; 24/04682-3 - Data analysis for automated harvest estimation in coffee plantations, BP.TT

Abstract

For the coffee grower to be able to practice good business with the sale of his coffee, some pre-harvest steps are key, among them, having a good estimate of production in the fields during the harvest, before starting the harvest. This step is crucial for the coffee grower to be able to optimize and plan the management of the producing areas and finances of the farms. In the past harvest, more than 30 million bags of Arabica coffee were produced in the country in more than 1.4 million hectares; therefore, the search for new technologies aimed at improving the coffee production sector is particularly interesting. In this sense, a feasibility study of a digital tool was proposed, with image analysis through artificial intelligence to predict the estimated harvest. The promising results presented in Phase 1 of this project (PIPE-FAPESP n 2020/05864-7, completed in September/2021) in the identification of grains and maturation level, which can be found in the phase 1 report attached to this proposal, justified the Project continuity for Phase 2. During Phase 2, which started in October/2022, it was proposed that this model and tool be incorporated into an accessible and intuitive digital platform, and strategically tested with a commercial objective. About 6 months after the beginning of Phase 2 execution, AgroBee received a contribution from a private investor in the amount of R$ 1 million, to help in the development of its technology and expand its commercial capacity and expression in the market. Thus, in accordance with the FAPESP norm, it is proposed to carry out this project in Phase Invest, to leverage the development of this digital tool, mainly in the refinement of the algorithm through a larger sample number - spatial and temporal - and also by automating inputs, offering greater security and practicality in the execution of the crop estimation tool. Ten farms located in the main arabica coffee producing regions will be monitored during the project years in three different moments of the coffee production cycle, (i) beginning of fruiting, (ii) advanced fruiting, and (iii) medium maturation. The data collected will allow refining the artificial intelligence tool and integrating it into a platform to be developed for validation and later commercialization. (AU)