CLASSIFICATION OF DATE FRUITS USING DEEP LEARNING MODEL

Abstract

Fruits are the most widely grown of all farm products worldwide and have the characteristic of possessing many varieties that differ from each other on the basis of their external characteristics such as color, length, diameter, and shape. These external attributes are equally critical in the definition of fruit types. However, identifying varieties of fruits using outside appearance only could require professional knowledge, and hence may consume lots of time and effort. The focus in the current study is to recognize various types of date fruits using artificial intelligence techniques. The varieties targeted are: Ajwa, Galaxy, Medjool, Meneifi, Nabtat Ali, Rutab, Shaishe, Sokari, and Sugaey. A total dataset of 1,658 images of the mentioned varieties was used in this research, which was obtained by means of a computer vision system (CVS). In contrast to conventional hand-crafted feature extraction-based work on color and shape, in this research study, a Convolutional Neural Network (CNN) model was employed for the direct classification of the images. The model indicated a very high accuracy of 98.8%, which validates the effectiveness and appropriateness of deep neural networks in classifying varieties of date fruit accurately.