GrapeNet: An Image Dataset for Grape Variety and Quality Classification.
Description
GrapeNet: A High-Resolution Image Dataset for Grape Variety and Quality Classification The GrapeNet dataset addresses a pressing need in precision viticulture for accurate, automated classification of grape varieties and quality levels. It includes 25,425 high-resolution images of grape clusters spanning three commercially important varieties: Black, Flame, and Green. These are further divided into eight subcategories based on freshness (fresh or rotten) and shape (round or elongated). The dataset was captured under consistent laboratory conditions with variations in lighting and camera angles to reflect real-world diversity, making it well-suited for machine learning applications. GrapeNet supports a wide range of computer vision tasks in agriculture, including automated fruit sorting, ripeness detection, quality control, and condition-based grading. The dataset is particularly valuable for researchers and developers working on deep learning-based classification and object detection models in agricultural contexts. Dataset Features: Total Images: 25,425 grape cluster images Categories: 3 varieties (Black, Flame, Green) Subcategories: 8 (based on freshness and shape) Black/Round Fresh – 2,900 Black/Round Rotten – 3,000 Flame/Round Fresh – 3,350 Flame/Round Rotten – 2,975 Green/Elongated Fresh – 3,500 Green/Elongated Rotten – 3,050 Green/Round Fresh – 3,675 Green/Round Rotten – 2,975 Image Resolution: Original 4624×3472 pixels, resized upto 3000×3000 pixels for modeling Format: JPEG (.jpg) Device Used: Xiaomi Redmi Note 13 5G (50 MP, f/1.79 aperture) Data Collection Methodology: Images were captured using smartphone cameras in a controlled indoor environment. Grapes were arranged on plain white or black backgrounds and photographed under both natural and artificial light. This approach ensured consistent imaging while preserving variability relevant to real-world use cases. To enrich the dataset further, image augmentation techniques such as flipping, rotation, and brightness adjustments were applied using Python scripts. Applications: Grape variety recognition Ripeness detection Rotten/fresh classification Smart sorting systems in vineyards and packing units Educational tools for AI in agriculture
Files
Steps to reproduce
Images were captured using smartphone cameras in a controlled indoor environment. Grapes were arranged on plain white background and photographed under natural light. This approach ensured consistent imaging while preserving variability relevant to real-world use cases. To enrich the dataset further, image augmentation techniques such as flipping, rotation, and brightness adjustments were applied using Python scripts.