Development of a system for classifying coffee beans by color, morphology and chemical characteristics

• Grant number: 22/12775-6
• Support Opportunities: Research Grants - Innovative Research in Small Business - PIPE
• Duration: February 01, 2023 - January 31, 2025
• Field of knowledge: Agronomical Sciences - Agricultural Engineering - Process Engineering of Agricultural Products
• Principal Investigator: Carlos Fernando Baltieri
• Grantee: Carlos Fernando Baltieri
• Host Company: Baltiere Indústria, Comércio, Exportação e Importação Ltda
• CNAE: Atividades de apoio à agricultura ; Fabricação de produtos à base de café
• City: Ribeirão Preto
• Associated researchers: Gustavo Trindade Valio ; Wanessa Melchert Mattos ; Winston Pinheiro Claro Gomes
• Associated scholarship(s): 23/09196-7 - Development of software for analysis and classification of color, morphology, and chemical characteristics of coffee beans, BP.TT; 23/06084-3 - Selection and application of suitable wavelengths for quantification of specific chemical characteristics in coffee beans, BP.TT; 23/02644-4 - Assistance for the evaluation and application of wavelengths outside the visible spectrum in the classification of chemical characteristics of coffee beans, BP.TT; 23/02574-6 - Development of a system for classifying coffee beans by color, morphology and chemical characteristics, BP.PIPE

Abstract

The present project aims to solve the existing challenges in the classification of coffee beans by carrying out a research and development of a specialized system for this purpose. The classification process of coffee bean is necessary due to the impossibility of the selectors equipment carry out a precise evaluation of the selected beans. This low precision is due to the high speed of the picker to make the decision (<2ms). Classification is performed manually, with a certain degree of subjectivity, demand for a highly qualified professional and is time consuming. It generates a bottleneck in the production process due to the damming of samples to be evaluated for batch release, in addition to reprocessing entire batches due to the impossibility of quality evaluating of selected bean during the production process. This manual classification can occur with a period up to 56 hours from the completion of the batch selection process (weekends). Due to these difficulties, the proposed project aims to carry out the research and development of a system (image capture equipment and evaluation software) for classifying coffee beans quickly, accurately and reproducibly. This classification will be carried out using static images of the two sides of the samples, evaluating the color (pixel by pixel that make up the grain - it is already in the MVP test phase) and morphology. In addition to these evaluations, which are the same performed by the classifiers, the evaluation of some chemical characteristics of the grains such as moisture content, fat content, fungal attack, etc., using wavelengths outside the visible spectrum will also be carried out. These characteristics are also subjectively evaluated through solution tasting, which is a more time-consuming process than visual classification due to the need to roast the beans and prepare the solution. In order to obtain such evaluations, this project will carry out research and development of improvements in the image capture equipment and in the modules for morphological evaluation and interest chemical characteristics. With this solution, the bottleneck generated by the classification will be extremely minimized and will give more freedom to the classifier and to the operation. In addition to enabling a faster classification, the production process will be able to carry out periodic evaluations during the selection process, obtaining data to adjust the selector during the process, avoiding rework. The target customers of this system are, initially, the classifiers, giving greater speed to their work. Also possible are the selection process team, giving greater autonomy, reducing rework, and farmers and buyers, allowing a better evaluation of the marketed product. (AU)