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

Jarin Tasmim Jinia, Md. Sakibur Rahman, Mayen Uddin Mojumdar, Md. Hasan Imam Bijoy

Abstract: This study highlights the growing significance of flowers, especially roses, in the global agricultural market, where they are cultivated for both personal enjoyment and commercial purposes. Among these, roses are considered one of the most popular and widely cultivated flowers. However, rose cultivators often encounter substantial challenges due to diseases that affect the plants, which can lead to significant economic losses in the agricultural sector. Timely and accurate detection of these diseases is crucial to mitigating their impact, potentially saving millions of dollars in crop losses. The dataset utilized in this research consists of 10,000 high-quality images collected from an initial set of 3113 images taken from several rose gardens located in Amin Model Town, Khagan, Ashulia, and Savar, Bangladesh. The data collection process spanned from October 30 to November 6, 2024. These images are categorized into four distinct classes: Healthy Leaf, Black Spot, Leaf Hole, and Dry Leaf, representing various stages of disease development in rose plants. The images were captured using a Vivo IQOO Z9x phone, ensuring high resolution and detailed imagery necessary for research analysis. This dataset serves as a valuable resource for researchers and developers working on creating efficient algorithms for the early and accurate identification of rose leaf diseases. By leveraging machine learning and image processing techniques, these algorithms could significantly enhance disease detection and prevention, helping to safeguard crops and reduce economic losses in the agricultural sector.

External IDs:doi:10.1016/j.dib.2025.112108