Reliability in visual evaluation of white and yellow corn grains using digital images in Venezuela

Abstract

The quality of the corn grain is a fundamental aspect during its processing and consumption, so its analysis must be very reliable. The objective of the study was to evaluate the reliability of the laboratory analysis of the corn quality by the grain appearance through digital images. For this, two hundred fractions of white and yellow corn grains were captured with high resolution scanner. The images were evaluated by 29 analysts with experiences ranging from less than 1 year to 25 years. The analysts classified the grain images in different categories based on Venezuelan regulations for industrial corn use. The results were analyzed using statistic parameters of mode (more frequent definition), frequency, the percentage of successes with the mode value of each analyst, and the kappa coefficient (general, experience of analyst, and intralaboratory groups). The data showed general kappa value of 0.42 (moderated) and 0.40 (discrete) for white and yellow corn, respectively. These coefficients varied among the intralaboratory groups and were not consistently associated with their years of experience. The majority of analysts showed achievements with the mode value ranging from 50 to 75 % for both corn types. These results expose the need to continue standardizing the criteria of evaluation to improve the reliability in the determination of the appearance of grains in white and yellow maize in Venezuela.