Algoritmo de clasificación visual de tomates chonto según norma NTC-1103-1 (parámetros de color, tamaño y forma)

Abstract

This article introduces the development and implementation a low-cost system for the classification of Chonto tomatoes according to their color, shape and size, in accordance with the Colombian technical standard NTC 1103-1. To achieve the proposed objective, a classification algorithm is developed using Python and OpenCV software. The obtained results show that the classification for color and maturity had an accuracy of 93%. In the classification by size the precision was 98%. Regarding the evaluation of eccentricity for determining the shape, an accuracy of 80% was obtained. The aforementioned precision values are comparisons with the result obtained manually by a trained person, which is considered the ideal classification. However, the response time of the algorithm is 0.48 sec, in average, which is much less than the time required for human inspection and classification. In addition to precision achieved, it can be said that the developed software responds to the need to detect and classify chonto tomatoes according to their color, size and shape established in the Colombian technical standard NTC 1103- 1.