The aim of this work was to improve the classification of green asparagus in an agro-export company by way of a Computer Vision System (CVS). Thus, an image analysis application was developed in the MATLAB® environment to classify green asparagus according to the absence of white spots and the width of the product. The CVS performance was compared with a manual classification using the error in the classification as the quality indicator; the yield from the raw material (%) and line productivity (kg/h) as the production indicators; and the net present value (USD) and internal rate of return (%) as the economic indicators. The CVS classified the green asparagus with 2% error; improved the yield from the raw material from 43% to 45%, and line productivity from 5 to 10 kg/h; and increased the net present value by 102,609.00 USD, yielding an Internal Rate of Return of 156.3%, much higher than the Opportunity Cost of the Capital (8.6%). Hence the classification of green asparagus by a CVS is an efficient and profitable alternative to manual classification.

Johonathan Salazar-Campos, Orlando Salazar-Campos, Osmar Gálvez-Ruiz, Herlita Gavidia-Chávez, Mery Gavidia-Chávez, Lorena Irigoin-Guevara, and Jesús Obregón-Domínguez. Prev Nutr Food Sci 2023;28:444-52. https://doi.org/10.3746/pnf.2023.28.4.444

In recent times, using image-based deep learning techniques has been very useful for classifying fruits. However, mango is a particular challenge because its shelf life is directly related to the degree of maturity at the time of harvest, which is determined mainly through the visual assessment of the operators when sorting them, doing it incorrectly, coupled with the blows and compressions exposed during storage and transport, trigger the deterioration in their quality, hence the importance of this operation. In this sense, the purpose of this study focuses on designing a computational model based on Deep Learning that uses a convolutional neural network (CNN) parameterized to select mangoes according to their color tone based on their maturity stage. A dataset was established from 201 labeled images, then a CNN architecture was built, obtaining a model on which 3 criteria were evaluated: the number of convolution and grouping layers, the resolution of the input images and the number of training epochs required. As a result, an input image resolution of 32 x 32 pixels was achieved, the process included a convolution layer followed by pooling that fed another layer that processed the results and ended in an output layer with 2 neurons under the One Hot Encoding representation. The model required 20 epochs to perfect its training, achieving 96.04 % accuracy in mango image selection. This study supports the potential of CNNs in quality analysis through mango color selection, even in uncertainty scenarios.