Grapes leaf disease dataset for precision agriculture.

Area of Science:

• Agricultural Science
• Computer Vision
• Machine Learning

Background:

• Grape cultivation is vital globally, but fungal diseases significantly reduce yield and quality.
• Accurate and timely disease identification is crucial for effective crop management and economic viability.
• Existing methods for grape disease detection may lack the scale and precision needed for advanced AI applications.

Purpose of the Study:

• To introduce a comprehensive, high-quality dataset of annotated grape leaf images for disease detection.
• To facilitate the development of advanced AI models for automated identification and classification of grape leaf diseases.
• To provide an open-access resource for researchers in computer vision and agricultural technology.

Main Methods:

• Collection of 2,726 high-resolution grape leaf images over two years (2023-2025) in Nashik, India.
• Expert annotation and organization of images into 'Healthy' and 'Unhealthy' categories, with unhealthy leaves further classified into Downy Mildew, Powdery Mildew, and Bacterial Leaf Spot.
• Validation of the dataset using a transfer learning approach with the ResNet-18 algorithm.

Main Results:

• The dataset comprises 2,726 meticulously annotated images of grape leaves.
• The ResNet-18 model achieved a 96% classification accuracy, demonstrating the dataset's high quality and suitability for deep learning.
• The dataset is structured into healthy leaves and three major disease categories: Downy Mildew, Powdery Mildew, and Bacterial Leaf Spot.

Conclusions:

• The developed grape leaf disease image dataset is a valuable resource for advancing AI-driven disease detection in agriculture.
• The high classification accuracy achieved validates the dataset's suitability for training robust machine learning models.
• This open-access dataset will support research and development for sustainable grape production and improved disease management strategies.