A comprehensive dataset of rice leaf images for disease detection using machine learning

Area of Science:

• Agricultural Science
• Computer Vision
• Plant Pathology

Background:

• Rice is a staple food for a significant portion of the global population.
• Accurate and timely detection of rice diseases is crucial for maintaining crop yields and ensuring food security.
• Existing datasets may not fully represent the diversity of real-world conditions and disease variations.

Purpose of the Study:

• To introduce a comprehensive, expert-annotated dataset of rice leaf images.
• To facilitate the development of machine learning models for automated rice disease identification.
• To support sustainable agricultural practices through early disease detection.

Main Methods:

• Collected 2,753 original rice leaf images from BRRI, capturing diverse environmental conditions.
• Applied data augmentation techniques (rotation, scaling, brightness, flipping) to create 16,247 augmented images.
• Ensured expert annotation by agronomy specialists for accurate disease classification across seven categories.

Main Results:

• A dataset of 19,000 images (2,753 original, 16,247 augmented) covering seven disease classes and healthy samples.
• Images captured using smartphone cameras reflect practical, real-world scenarios.
• Expert annotations provide reliable ground truth for model training and validation.

Conclusions:

• The dataset is a valuable resource for training robust AI models for rice disease detection.
• Early disease detection can significantly improve crop management and reduce yield losses.
• This work contributes to enhancing food security in rice-dependent regions through technological advancement.