The process of selecting specialty coffee beans requires three types of supervision. Two of them are physical and include raw and roasted coffee samples. The third is sensory and involves tasting the beverage. Certification is provided by the Specialty Coffee Association of America (SCAA).
According to SCAA guidelines, coffee quality is measured on a decimal scale from zero to 100. A specialty coffee must score 80 or more. Before making a report, the grower sends a sample of raw beans from each batch to three cups (tasters) that make and roast coffee, again in accordance with SCAA standards.
But they are collaborating with Brazilian scientists at the Center for Nuclear Energy in Agriculture at the University of São Paulo (CENA-USP), colleagues at the Luiz de Queiroz Faculty of Agriculture (ESALQ-USP), and the Computer Center at the Federal University of Pernambuco (UFPE). developed a coffee bean selection method based on multispectral imaging and machine learning. The method does not require roasting and can be performed in real time during the production process. Although it is based on expensive equipment, it avoids possible human errors. A article recently published about the new method Computer and Electronics in Agriculture.
“Specialty coffees are often selectively harvested, meaning only ripe red cherries are picked. They are individually hand-harvested. If a specialty coffee grower harvests green beans or uses sliver picking at any time, manually and/or mechanically, this procedure is comparable to a standard commercial product. may result,” says Winston Pinheiro Claro Gomes, first author of the paper. Gomes is a PhD candidate in chemistry at CENA-USP. Vanessa Melchert Mattos and Clisia Barboza da Silva as thesis advisors.
“In our method, we separate custom and standard commercially accepted beans using a combination of multispectral imaging and mathematical algorithms that process the data provided by the images,” said Gomes. “Specialty coffee should score between 80 and 100, but our model cannot tell whether the beans are 80 or 90. This requires machine learning with samples for each score to identify these categories in the mathematical model.”
The research was conducted with the support of the USA. Young Researcher Scholarship given to Barboza da Silva, the penultimate author of the article, and Regular Research Scholarshipawarded to Mattos, the last author of the article.
multispectral methodology
The team used a multispectral imaging (MSI) technique based on reflection and autofluorescence, in which images of the same object are taken at different wavelengths, and then followed a machine learning model to classify the beans based on the information gathered from the images.
“MSI’s use in the coffee industry is very new. “As can be seen from the literature on the subject, it is mostly used to map nitrogen in coffee gardens, detect necrosis in beans, and detect pests and diseases in plants.”
Researchers analyzed 16 samples of green beans from specialty and standard commercial crops grown in the states of Minas Gerais and São Paulo. Ten of the specialty coffee beans (coffee arabic) was from the 2016/17 crop grown in the Alta Mogiana region. They were rated in the 2017 Alta Mogiana Coffee Competition and were supplied by the region’s specialty coffee producers’ association. The other six samples were from standard commercial products purchased in bulk from the local market.
For each sample, 64 untreated cores were randomly separated, yielding a total of 1,024 (384 standards, 640 custom) cores and used for machine learning calibration, validation, and testing.
Gomes summarized the procedure: “We placed the beans in a Petri dish and put them in the device, which is a sphere containing LEDs, optical filters, and a camera. The camera was lowered until the samples were completely covered and captured images of different wavelengths after homogeneous and diffuse illumination. It first took monochrome reflection images and then autofluorescence images, then information on regions of interest was extracted by the firmware and used to build algorithms that classify the samples and give us the results.”
Principal component analysis (PCA) was then performed to explore the variables that influence the differences between specialty and standard coffees. The researchers ran four machine learning algorithms with a support vector machine (SVM) that proved the best and was used to calculate the coefficients to estimate the underlying variables.
fluorescent
In the visible spectrum (RGB) images, it was observed that the special nuclei were more uniform in shape, while the standard nuclei were more dense in the autofluorescent images. “Our mathematical model and algorithms use signal intensity information from fluorescence images. Some compounds in beans may be more excited at a particular wavelength. A more or less intense fluorescence signal may also be related to the change in concentration of a compound in beans, for example. and standard coffee beans.In this model, the most important information for establishing separation boundaries came from green fluorescence.Therefore, we decided to analyze individual compounds that naturally fluoresce green, and some fluorescent compounds that may affect the coffee grading separation process. We tried to relate it,” said Gomes.
Green fluorescence, a biological marker represented by green light in the visible spectrum, was analyzed for 10 phenolic compounds, and the data showed that catechin, caffeine, and some acids (4-hydroxybenzoic acid, sinapic acid, and chlorogenic acid) responded intensely after stimulation. It emits energy at 500 nm, with blue light at 405 nanometers (nm). These autofluorescence data (excitation/emission at 405/500 nm) contributed most to distinguishing green specialty beans from green standard beans.
“These are chemical species associated with aromatic groups that absorb energy of a particular wavelength. In autofluorescence-based methods, differences in the levels of these chemical species in specialty and standard coffee grades can be used to distinguish the two groups,” he said.
Differences in the levels of these compounds are typically used to distinguish between specialty and standard coffee beans. “For my graduate research, I studied the chemical composition of these samples and although there were no differences in the chemical species, we found differences in their concentrations, specifically the levels of chlorogenic acid and caffeine,” he said.
The next steps, according to Gomes, will require taking samples from each of the SCAA-defined score levels for specialty coffees (not an easy task) and grading the beans according to their scores. “In Brazil, coffees score 90-92 at most. It’s hard to find higher than that. For example, only coffee imported from Ethiopia gets 100 points. In my PhD research, I am trying to classify beans on the basis of X-ray images and I decided to increase the number of samples and the breadth of coffee. We analyze by including imported beans,” he said.
About the São Paulo Research Foundation (FAPESP)
The São Paulo Research Foundation (FAPESP) is a public institution with a mission to support scientific research in all fields of knowledge by providing fellowships, fellowships and grants to researchers affiliated with higher education and research institutions in the State of São Paulo, Brazil. FAPESP is aware that the best research can only be done by working with the best international researchers. It has therefore established partnerships with funding agencies, higher education, private companies and research organizations in other countries known for the quality of their research, and encourages scientists funded by their grants to further develop international cooperation. Find out more about FAPESP at www.fapesp.br/en and visit the FAPESP news agency at www.agencia.fapesp.br/en to stay up to date with the latest scientific breakthroughs that FAPESP has helped achieve through its many programs, awards and research centres. you can. You can also subscribe to the FAPESP news agency at http://agencia.fapesp.br/subscribe.
DAILY
Computer and Electronics in Agriculture
ARTICLE TITLE
Multispectral imaging application with machine learning models to distinguish between specialty and traditional green coffee
ARTICLE PUBLICATION DATE
2-June-2022