Evaluating the Classification of Freeze-Dried Slices and Cubes of Red-Fleshed Apple Genotypes Using Image Textures, Color Parameters, and Machine Learning

Dried red-fleshed apples are considered a promising high-quality product from the functional foods category. The objective of this study was to compare flesh features freeze-dried belonging ‘Alex Red’, ‘Trinity’, ‘314’, and ‘602’ genotypes indicate which parameters shapes dried samples most useful distinguish apple genotypes. Apple were at stage harvest maturity. average fruit weight, starch index, internal ethylene concentration, firmness, total soluble sugar content, titratable acidity determined. One hundred slices with thickness 4 mm one cubes dimensions 1.5 cm × each genotype subjected freeze-drying. For sample (slice or cube), 2172 image texture extracted images in 12 color channels, L*, a*, b* classification models developed based on set selected textures combined using various traditional machine-learning algorithms. Models built slice 11 channels correctly classified an overall accuracy reaching 90.25% mean absolute error 0.0545; by adding (L*, b*) models, increase 91.25% decrease 0.0486 observed. cube including characterized up 74.74%; resulted 80.50%. greatest mixing cases observed between Red’ ‘Trinity’ as well ‘314’ cubes. can be used practice non-destructive manner. It avoid different chemical properties. Further studies focus deep learning addition machine build samples. Moreover, other drying techniques applied, predict changes structure estimate properties