Contributing to agriculture by using soybean seed data from the tetrazolium test.

Abstract

Agribusiness has a great relevance in the world׳s economy. It generates a considerable impact in the gross national product of several nations. Hence, it is the major driver of many national economies. Nowadays, from each new planting to harvesting process it is mandatory and crucial to apply some kind of technology to optimize a given singular process, or even the entire cropping chain. For instance, digital image analysis joined with machine learning methods can be applied to obtain and guarantee a higher quality of the harvest, leading to not only a greater profit for producers, but also better products with lower cost to the final consumers. Thus, to provide this possibility this work describes a visual feature dataset from soybean seed images obtained from the tetrazolium test. This is a test capable to define how healthy a given seed is (e.g. how much the plant will produce, or if it is resistant to inclement weather, among others). To answer these questions we proposed this dataset which is the cornerstone to provide an effective classification of the soybean seed vigor (i.e. an extremely tiresome human visual inspection process). Besides, as one of the most prominent international commodity, the soybean production must follow rigid quality control process to be part of world trade. Hence, small mistakes in the seed vigor definition of a given seed lot can lead to huge losses.

Code and data availability

The paper describes a public visual-feature dataset from soybean seed tetrazolium test images, explicitly deposited on GitHub by the authors.