Work

A machine learning approach for classifying date fruit varieties at the Rutab stage

Year: 2025

Type: article

Abstract: Introduction: Dates have long been a vital part of the cultural and nutritional heritage of arid regions, particularly in the Middle East. Among their ripening stages, the Rutab stage-an intermediate phase between the Khalal (immature) and Tamar (fully ripe) stages-holds unique significance in terms of taste, texture, and market value. However, the classification of Rutab varieties remains underrepresented in the literature. Methods: To address this gap, we present a pipeline that leverages machine learning to classify Rutab dates from images. A custom dataset comprising 1,659 images across eight popular Rutab types was collected, and several deep learning models were evaluated. Results and discussion: Among the tested models, YOLOv12 achieved the highest recall of 93%. The proposed system is deployed within a mobile application, aiming to promote cultural preservation and increase global awareness of the diversity found within date varieties.

Source: Frontiers in Plant Science

DOI: 10.3389/fpls.2025.1678757