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

Description

Date fruits (Phoenix dactylifera L.) are among the most economically and nutritionally important fruits cultivated worldwide, particularly in the Middle East, North Africa, and South Asia. Different cultivars exhibit distinct characteristics in color, texture, shape, size, taste, and market value, making accurate identification essential for quality control, commercial trading, consumer confidence, and agricultural research. Traditional cultivar identification relies on visual inspection by experts, a process that is often subjective, time-consuming, and prone to human error. Consequently, there is growing interest in automated classification systems based on computer vision and deep learning. To support research in this domain, we present a comprehensive multi-variety date fruit dataset designed for cultivar classification and intelligent agriculture applications. The dataset facilitates the development and evaluation of machine learning, deep learning, transfer learning, and vision transformer models for automated date fruit recognition. By providing a diverse collection of labeled images, it serves as a benchmark resource for researchers and developers working in agricultural technology and computer vision. The original images were collected manually using a Samsung Galaxy A52s (5G) smartphone equipped with a 16 MP camera. Data acquisition was conducted at multiple local fruit markets to capture realistic consumer-level conditions and natural variations in fruit appearance inside Dhaka division. The original dataset consists of 1,865 JPG images belonging to seven date fruit cultivars. Most images were captured at a resolution of 4624 × 3468 pixels under natural daylight conditions. While three images has different resolution due to camera settings. The dataset contains seven date fruit cultivars: AJWA, CHHARA, DABBAS, KHURMA, MARYAM, SUKKARI, and ZAHEDI. Each image was carefully reviewed and labeled according to its corresponding cultivar to ensure reliable annotations. The original dataset consists of 1,865 JPG(.jpg) images distributed across the seven classes: AJWA (254), CHHARA (249), DABBAS (265), KHURMA (294), MARYAM (265), SUKKARI (268), and ZAHEDI (270). To increase dataset diversity and improve model generalization, several lightweight augmentation techniques were applied, including small rotations, brightness adjustments, contrast enhancement, zooming, horizontal flipping, and slight Gaussian blurring. These augmentations simulate common variations encountered during image acquisition while preserving the intrinsic characteristics of each cultivar. The augmented dataset contains 26,110 JPG(.jpg) images, comprising AJWA (3,556), CHHARA (3,486), DABBAS (3,710), KHURMA (4,116),MARYAM (3,710), SUKKARI (3,752), and ZAHEDI (3,780) images. By combining real-world image acquisition, reliable annotations, and data augmentation, it contributes to the development of robust AI-driven solutions for the date fruit industry.

Steps to reproduce

Data Collection: Collect date fruit samples representing seven cultivars: AJWA, CHHARA, DABBAS, KHURMA, MARYAM, SUKKARI, and ZAHEDI. Data Annotation: Organize images into cultivar-specific folders. Manually verify and label each image according to its corresponding date fruit cultivar. Remove duplicate, low-quality, or severely blurred images. Data Augmentation Techniques: To increase dataset diversity and improve model generalization, the following augmentations were applied: Small Rotation ( ±15 degrees). Brightness Adjustment. Contrast Enhancement. Zooming. Horizontal Flip. Slight Gaussian Blur. avoid heavy augmentation. Image Format: Format: JPG (.jpg) Color Mode: RGB This dataset is suitable for: Date fruit cultivar classification Deep learning and transfer learning research Vision Transformer (ViT) applications Computer vision benchmarking Precision agriculture research Agricultural product quality assessment Academic and student research projects

Institutions

• Daffodil International University

  Dhaka Division, Dhaka

Categories

Licence