RiceLeafDiseaseBD: A Field-Based Annotated Smartphone Image Dataset of Healthy and Diseased Rice Leaves from Bangladesh

RiceLeafDiseaseBD is a large-scale rice leaf disease dataset containing 9,769 high-resolution images (1024×1024 pixels) with 74,836 bounding box annotations. Images were captured using smartphones (CMF Phone 2 Pro and Google Pixel 5) from rice fields in Gazipur, Bangladesh, during the Summer 2025 growing season.

Dataset Contents:
6 classes: Healthy, Blast, Brown Spot, Leaf Smut, Rice Tungro, Sheath Blight
Original images organized by class folders
YOLO format annotation files for object detection
CSV metadata file (image IDs, original size, resize size)

Key Features:
Field-based collection with natural backgrounds
Focus on BRRI dhan28 and BRRI dhan29 varieties
Expert-validated annotations by plant pathologists, farmers
Compatible with TensorFlow, PyTorch, YOLOv5/v8

For detailed information, see the README file included in the dataset.