A Comprehensive Study of Machine Learning Algorithms for Date Fruit Genotype Classification
M. Venkata Subbarao, Usha Rani, J T S Sindhu, Gaurav Kumar, Viswanadham Ravuri, N Silpa
Abstract
There are over 200 distinct date fruit varieties around the globe. Physical characteristics such as size and structure (collectively known as morphological attributes), colour, and other shape characteristics are used to manually determine their class of genome type. Date fruit processing industries require expert systems to determine the type of a fruit based on its exterior without a great deal of effort and time-consuming information. This study aims to identify the type of date fruit among the seven varieties Sukkary, Ruthana, Barhee, Safawi, Deglet Nour, Sagai, and Rotab Mozafati, which are popularly cultivated in Turkey and other Middle Eastern countries. Using image processing techniques, a total of 34 attributes were extracted from each date fruit, including morphological (12), shape (4), and colour (18). Performance analysis of machine learning models is investigated with the different sets of extracted features to know the superior classifier. Analysis is carried out with individual sets of features and combinations of features.