2019 ASHS Annual Conference

Improving and grading for banana quality has become an important requirement to increase fruit exportation. At present, during postharvest processing, banana fruit is sorted and classified manually according to maturity as indicated by the peel color. This manual grading is a destructive method and labor-intensive. The objective of this study was to determine the efficiency of computer vision (CV) system with RGB color camera to evaluate the maturity of banana as indicated by the color of the peel and to predict the internal chemical characteristics, primarily soluble solid contents and starch. Three cultivars of banana (‘Cavendish’, ‘Malindi’ and ‘Milk banana’) were obtained from the local market. At every stage (from stage 2 to stage 7 of banana ripening), the samples were imaged individually using a color RGB camera. A software (ImageJ) was used to determine RGB color type and intensity. Whereas, pocket refractometer was used to measure total soluble solids (TSS). Results showed that CV can predict the amount of TSS and starch in the fruit. Hue and a/b color indices were more accurate to detect the amount of TSS and starch for all cultivars. As a result, a model for all three cultivars can be created from the data of all ripening stages (green to over ripe stages) of banana to predicate TSS and starch of sample from the skin color of banana.