Red-flesh kiwifruit inner quality scoring by a computer vision system

Red-fleshed kiwifruit have recently entered the international market. As their nutraceutical properties have generated significant consumer interest, to establish a quality standard along the supply chain, the internal quality of the fruit must be accurately evaluated. Evaluating the redness of kiwifruit flesh poses a complex challenge due to the inherent variability in color localization and spottiness, as well as the wide range of red shades and intensities within each fruit. The current evaluation protocol employs experienced human raters to visually assess the internal quality of red-fleshed kiwifruit. Quality assessment requires the evaluator to score the red quantity and both inner and outer pericarp red intensities to determine the fruit category. As a consequence, this method suffers from complexity, subjectivity, limited repeatability, and is a slow process. In this study, computer vision and unsupervised learning algorithms were used to develop a computer vision system capable of scoring fruits according to both red quantity and red intensity. RGB images of equatorially sliced fruits were segmented into the Hue-Saturation-Value color space to generate the red shades mask (RSmask). The latter was used to extract quantity descriptors for the K-means classifier, namely "red quantity" classifier. Simultaneously, the RSmask was applied to segment a red-related image from which the red intensity descriptors were obtained through a matrix conversion to the CIELAB color space. Similarly, the "red intensity” classifier was used to score the red intensity of the flesh. A total of 102 sample fruits were classified into 36 categories based on the combination of the red quantity and intensity scores. The results demonstrated that red color quantity is much more predictable than red color intensity due to human eye color perception issues such as color constancy and simultaneous contrast.

Mirko Piani won the ISHS Young Minds Award for the best oral presentation at the II International Symposium on Precision Management of Orchards and Vineyards in Australia in December 2023.

Mirko Piani, University of Bologna, Viale Fanin 46, 40127 Bologna, Italy, e-mail: mirko.piani2@unibo.it

The article is available in Chronica Horticulturae