Artificial intelligence tools and a diagrammatic scale for evaluating the quality of coating in treated soybean seeds

Abstract

Image analysis combined with artificial intelligence has enabled efficient evaluation of the quality of seed lots during the production process. In the treatment stage, the evaluation of seed coverage is performed by visual analysis, which is considered subjective and of low standard. With the aim of helping producers and those responsible for treatment in classifying the coating of soybean seeds, the objective of this research was to investigate the efficiency of artificial intelligence tools for analyzing the quality of coating of soybean seeds submitted to chemical treatment and develop a diagrammatic scale for the standardization of the evaluation for this characteristic. To obtain the different levels of coating, excellent, good, medium and poor, images were captured and processed using the GroundEye® system, version S800, of samples of industrially treated soybean seeds (TSI) and treated on the farm (on-farm). The quantification of seed coverage was performed by adding the percentages of the red and pink color dominance characteristics detected in the seeds. Subsequently, artificial intelligence tools, classifiers and a decision network were used to help classify the coating quality. From the results obtained and considering the maximum and minimum values of quantification of the coating observed in the image bank obtained from the samples, a diagrammatic scale was established for the visual evaluation of the quantity and quality of the coating of the seeds with the chemical treatment. The use of artificial intelligence tools was efficient for evaluating the quality of soybean seed coatings.