Classification of tomato ripeness in the agricultural industry using a computer vision system

Abstract

Machine vision systems (SVA) occupy an important place in the field of food and agriculture, these techniques are performed in situ, are efficient, non-invasive, time-saving and more economical than traditional techniques. Tomatoes (Solanum lycopersicum) are extensively cultivated throughout the world, are essential in the agricultural and culinary fields and are recognized for their beneficial contributions to health. Lack of knowledge about fruit maturity, proper harvesting and postharvest handling are factors responsible for large postharvest losses. Therefore, the objective of this research was the construction of a VAS that allows establishing relationships between color and maturity stage of the Chonto Roble F1 tomato. The VAS built is composed of hardware and software duly synchronized through the application of computer vision algorithms in Python 3.9 software that allow it to perform the acquisition and segmentation of the image and present the user with the color coordinates in the CIEL*a*b* system. Finally, color measurements were performed on tomato samples at different stages of ripening in the VAS and a HunterLab ColorQuest XE (EHL) spectrophotometer. The results obtained indicated that there are no significant differences in both measurement systems for L* values, the changes produced in b* and a* were statistically significant for tomato samples. The results obtained indicated the potential use of the constructed VAS for the determination of the degree of maturity of tomatoes in real time, in a non-invasive and low-cost way.