Notícia

Garretts Simpson.design

Technique based on artificial intelligence Enables automation of crop seed analysis

Publicado em 19 março 2021

In Brazil, researchers affiliated with the Center for Nuclear Energy in Agriculture (CENA) and the Luiz de Queiroz College of Agriculture (ESALQ), both part of the University of São Paulo (USP), have developed a methodology based on artificial intelligence to automate and enhance seed quality evaluation, a process required by law and currently done manually by analysts licensed with the Ministry of Agriculture.

The group used light-based technology such as that deployed in plant and cosmetics analysis to get images of the seeds. Then they turned to machine learning how to automate the image interpretation process, minimizing a few of the difficulties of traditional procedures. For example, for many species, optical imaging technologies may be used to an entire batch of seeds rather than samples, as is the case currently. Furthermore, the technique is non-invasive and doesn’t destroy the products analyzed or generate residues.

The light-based techniques consisted of chlorophyll fluorescence and multispectral imaging. Among plants that are applicable as both crops and experimental models, the researchers chose carrots and berries produced in various countries and seasons and filed to different storage requirements. They utilized seeds of the Gaucho and Tyna commercial tomato varieties made in Brazil and the US, and seeds of the Brasilia and Francine carrot types created in Brazil, Italy, and Chile.

The decision has been based on the economic relevance of the food crops, for which world demand is high and increasing, and on the difficulties faced by growers in amassing their seeds. In both tomatoes and carrots, the ripening process is not uniform because the plants flower always and seed production is non-synchronous, so that seed lots may contain a combination of immature and mature seeds. The presence of immature seeds isn’t readily detected by visual procedures, and methods based on machine vision can diminish this problem.

The researchers compared the outcomes of the non-destructive analysis with those of classic germination and vitality tests, which are harmful, time-consuming, and labor-intensive. In the germination test, seed analysts separate trials, sow them to germinate in favorable temperature, water, and oxygen requirements, and verify the last quantity of ordinary seedlings produced based on the rules established by the Ministry of Agriculture. Vigor tests are complementary and more sophisticated. The most common are determined by the seed’s response to stress and seedling growth parameters.

Besides the difficulties mentioned, traditional approaches are time-consuming. In the case of tomatoes and carrots, by way of instance, it can take up to two weeks to obtain results, which can also be mostly subjective, depending on the analyst’s interpretation. “Our proposal is to automate the process as much as possible using chlorophyll fluorescence and multispectral imaging to analyze seed quality. This will avoid all the usual bottlenecks,” explained Clíssia Barboza da Silva, a researcher in CENA-USP and one of the authors of the article on the study published in Frontiers in Plant Science.

Silva is the principal investigator for the project supported by São Paulo Research Foundation – FAPESP. The lead author of this guide is Patrícia Galletti, who conducted the study as part of her master’s study and won the Best Poster Award in 2019 in the 7th Seed Congress of the Americas, where she recently presented partial results of the project.

Chlorophyll as a mark of grade

Chlorophyll is found in seeds, where it provides energy to the storage of all nutrients needed for growth (lipids, proteins, and carbs ). Once it has fulfilled this function, the chlorophyll breaks down. “However, if the seed doesn’t complete the maturation process, this chlorophyll remains inside it. The less residual chlorophyll, the more advanced the maturation process and the more and higher-quality the nutrients in the seed. If there’s a lot of chlorophyll, the seed is immature and its quality is poor,” Silva said.

If lighting at a specific wavelength is shone on the chlorophyll in a seed, then it doesn’t move the energy to another molecule but instead re-emits the light at another wavelength, meaning that it fluoresces. This fluorescence can be measured, she explained. Red lighting may be utilized to excite chlorophyll and capture the fluorescence by means of a system that converts it into an electric signal, producing a picture comprising gray, black, and white pixels. The milder regions correspond to high levels of chlorophyll, indicating that the seed is unlikely to germinate.

Artificial intelligence

In multispectral imaging, LEDs (light-emitting diodes) emit light from the visible portion of the spectrum as well as non-visible light (UV and near-infrared). To analyze seed quality based on reflectance, the researchers utilized 19 wavelengths as well as the results with quality assessment data obtained by traditional methods. The best results were obtained with near-infrared in the instance of lettuce seeds and UV in the case of tomato seeds.

Seeds contain proteins, lipids and sugars that consume a portion of the light emitted by the LEDs and reflect the rest. The reflected light is captured by a multispectral camera, and the picture captured is processed to separate the seeds from the aid in the apparatus, which corresponds to black pixels without value, while the seeds are gray-scale. The values of these pixels in the image of a seed correspond to its chemical makeup.

“We don’t work with an average result for a sample. We perform individualized extraction for each seed,” Silva said. “The larger the amount of a given nutrient the seed contains, the more light of a specific wavelength it absorbs so that less is reflected. A seed with a smaller nutrient content contains fewer light-absorbing molecules. This means its reflectance is higher, although this varies according to its components, which behave differently depending on the light wavelength used.”

An algorithm describes the wavelength that obtains the best outcome. The process provides information about the seed’s chemical composition, from which its quality could be inferred.

For the researchers, it wasn’t sufficient to reach the imaging stage, since it is still an operation which requires human monitoring. “We then deployed chemometrics, a set of statistical and mathematical methods used to classify materials chemically,” Silva said. “The idea was that the equipment should classify quality on the basis of the image it captured.” The methods employed by the scientists in the research are frequently utilized in medicine as well as the food market.

Nextthey leveraged machine learning how to examine the versions made using chemometrics. “We taught the model to identify high-quality and low-quality seeds. We used 70% of our data to train the model, and used the remaining 30% for validation,” Silva said. Quality classification precision ranged from 86percent to 95% in the case of tomato seeds, and from 88% to 97percent in the case of carrot seeds.

The two chief techniques were equally accurate and time-saving, given the speed of image capture. The chlorophyll fluorescence tool captured one image a second, whereas the multispectral imaging analyzer processed 19 pictures in five moments.

Unexpected results

An unexpected result produced in the duration of the project proved highly significant. Chlorophyll fluorescence and multispectral imaging are also efficient techniques for plant variety screening, an essential portion of seed lot analysis to prevent economic losses. “Growers buy seeds with the expectation of a certain crop yield, but production will be affected if seeds with different genetic characteristics aren’t properly separated,” Silva said.

Screening is presently done by analysts trained in the skills needed to grade seeds by color, shape, and size, in addition to molecular markers where possible. From the analysis, the two techniques demonstrated efficient to separate lettuce varieties but multispectral imaging was unsatisfactory in the case of tomato types.

“The study produced novel results with regard to the use of fluorescence to screen varieties,” Silva said. “We found no prior research in which fluorescence was used for this purpose. Some studies show multispectral imaging to be efficient for this purpose, but not with the instrument we used.”

Instrument sharing

A fantastic method to move the knowledge produced by the study to the productive industry, Silva said, is to own companies develop the gear available to seed manufacturers. “It would be possible to use the results of our research to develop an instrument that used only UV light to characterize tomato seed quality and bring it to market, for example,” she surmised.

More info:

Patrícia A. Galletti et al, Integrating Optical Imaging Tools for Rapid and Non-invasive Characterization of Seed Quality: Tomato (Solanum lycopersicum L.) and Carrot (Daucus carota L.) as Study Cases, Frontiers in Plant Science (2020).

Citation:

Strategy based on artificial intelligence enables automation of harvest seed analysis (2021, March 19)

Recovered 19 March 2021

From https://phys.org/news/2021-03-technique-based-artificial-intelligence-automation. html

This document is subject to copyright. Apart from any fair dealing for the purpose of personal research or study, no
Part may be duplicated without the written permission. The information is provided for information purposes only.