Comprehensive Date Fruit (Phoenix dactylifera L.) Dataset for Cultivar Classification, Intelligent Agriculture Applications and Computer Vision Research

DateNet is a publicly available image dataset of date fruits (Phoenix dactylifera L.) developed to support research in computer vision, machine learning, deep learning, transfer learning, and intelligent agriculture. The dataset contains images of seven commercially available date fruit cultivars collected from local fruit markets within the Dhaka Division of Bangladesh under natural daylight conditions. The original dataset consists of 1,861 manually labeled JPG images captured using a Samsung Galaxy A52s (5G) smartphone equipped with a 16 MP camera. All images were acquired at a resolution of 4624 × 3468 pixels under natural daylight conditions. Images were acquired under real-world consumer-level conditions and include natural variations in illumination, orientation, scale, background, and fruit arrangement. All images were manually reviewed and assigned to their corresponding cultivar labels. The dataset contains seven date fruit cultivars: AJWA (252 images), CHHARA (249 images), DABBAS (265 images), KHURMA (294 images), MARYAM (263 images), SUKKARI (268 images), and ZAHEDI (270 images). This repository includes: 1_Original dataset.zip – Original manually labeled date fruit images organized by cultivar. Augmented_Training_Set.zip – Augmented training images generated using rotation, brightness adjustment, contrast enhancement, random scaling, horizontal flipping, and Gaussian blurring. DateNet_Metadata_Detailed.csv – Image-level metadata containing acquisition timestamps, device information, image dimensions, file sizes, and other image attributes. Augmentation_Log.csv – Detailed records of the augmentation procedures applied to training images. Augmented_Report.csv – Summary information regarding the augmented training dataset. The dataset is intended for applications including date fruit cultivar classification, image classification, transfer learning, vision transformers, explainable artificial intelligence (XAI), agricultural informatics, intelligent agriculture, automated fruit recognition, and food authentication systems. By providing original images, metadata, augmentation records, and training-ready augmented data, DateNet supports reproducible research and benchmarking of computer vision models for date fruit cultivar recognition.