Annotated image dataset of fire blight symptoms for object detection in orchards

Area of Science:

• Agricultural Science
• Plant Pathology
• Computer Vision
• Digital Agriculture

Background:

• Fire blight (Erwinia amylovora) poses a significant threat to fruit production, causing epidemic spread and substantial economic losses.
• Effective plant disease control necessitates enhanced spatial and temporal resolution in monitoring systems.
• Digital monitoring systems offer a promising approach to improve the efficiency of plant disease management in orchards.

Purpose of the Study:

• To develop a digital disease monitoring system for fire blight using RGB images.
• To create and present the ERWIAM dataset, a valuable resource for training and validating plant disease detection algorithms.
• To facilitate the early detection and containment of Erwinia amylovora infections in orchards.

Main Methods:

• Collection of RGB image data across nine dates under diverse weather conditions between 2021 and 2024.
• Acquisition of images from various apple genotypes, including cultivars, wild species, and breeding progeny, artificially inoculated with Erwinia amylovora.
• Manual annotation of fire blight symptoms (shoots, flowers, leaves) and other plant diseases using the Computer Vision Annotation Tool (CVAT) in YOLO 1.1 format.

Main Results:

• The ERWIAM dataset comprises 1611 manually labelled RGB images (1280x1280 pixels) of fire blight symptoms and 87 background images.
• Annotations include bounding boxes for fire blight symptoms at various developmental stages and other plant diseases.
• The dataset is structured for use in developing and testing digital plant disease monitoring systems.

Conclusions:

• The ERWIAM dataset provides a comprehensive resource for researchers and developers in the field of digital plant disease monitoring.
• This dataset can significantly contribute to the advancement of automated detection systems for fire blight and other plant diseases.
• Utilizing such datasets is crucial for improving early-stage infection detection and mitigating economic damage in fruit production.