RoseLeafInsight: A high-resolution image dataset for rose leaf disease recognition

Area of Science:

• Agricultural Science
• Computer Science
• Artificial Intelligence

Background:

• Rose cultivation is vital to Bangladesh's economy but threatened by leaf diseases, leading to reduced yields and farmer income.
• Limited farmer education and access to experts hinder early disease identification and intervention.
• Existing challenges include pesticide resistance and natural factors impacting rose production.

Purpose of the Study:

• To present a comprehensive rose leaf disease dataset for enhancing disease tracking, diagnosis, and research.
• To facilitate the development of high-accuracy machine learning models for early disease detection in roses.
• To support precision agriculture through AI-driven insights for improved crop management.

Main Methods:

• Conducted large-scale field surveys from October 2024 to January 2025 to collect high-quality images of rose leaves.
• Created a dataset of 3,228 images across four classes: Black Spot, Insect Hole, Yellow Mosaic Virus, and Healthy.
• Applied image pre-processing (resizing to 3000x3000) and augmentation techniques (rotation, flipping, etc.) to enhance dataset quality and model generalization.

Main Results:

• The dataset comprises 3,228 images: 409 Black Spot, 453 Insect Hole, 680 Yellow Mosaic Virus, and 1,686 Healthy.
• Initial testing with MobileNetV2 achieved 96.79% accuracy in classifying rose leaf diseases.
• The dataset serves as a benchmark for training deep learning models for automated monitoring.

Conclusions:

• The developed rose leaf disease dataset is crucial for advancing AI-driven disease detection in agriculture.
• This resource enables the creation of effective machine learning models for early diagnosis, empowering farmers with timely interventions.
• Integration with farming equipment like drones can enhance real-time monitoring and decision-making in precision agriculture, leading to smarter, more efficient farms.