RoseLeafSet: Real-world leaf image dataset for AI-based agricultural solutions

Area of Science:

• Agricultural Science
• Plant Pathology
• Computer Vision

Background:

• Roses are economically significant globally, but susceptible to diseases causing substantial crop loss.
• Accurate and timely disease detection is vital for mitigating economic impact in rose cultivation.
• Existing methods for rose disease identification face challenges in speed and accuracy.

Purpose of the Study:

• To develop and present a comprehensive dataset for training machine learning models for rose leaf disease identification.
• To facilitate research in automated early detection of rose plant diseases.
• To contribute to reducing economic losses in the rose agricultural sector.

Main Methods:

• Collected 10,000 high-resolution images of rose leaves from various locations in Bangladesh.
• Categorized images into four classes: Healthy Leaf, Black Spot, Leaf Hole, and Dry Leaf.
• Utilized a Vivo IQOO Z9x phone for image acquisition to ensure quality and detail.

Main Results:

• Established a diverse dataset of 10,000 images representing healthy and diseased rose leaves.
• The dataset includes specific disease indicators such as black spot, leaf hole, and dry leaf.
• Provided a valuable resource for developing and testing machine learning algorithms for plant disease detection.

Conclusions:

• The created dataset is a significant resource for advancing automated rose disease detection.
• Machine learning and image processing techniques applied to this dataset can improve early identification of diseases.
• Enhanced disease detection can lead to better crop management and reduced economic losses for rose cultivators.