A comprehensive dataset of mandarin leaf images for classification.

Area of Science:

• Agricultural Science
• Computer Science

Background:

• Citrus cultivation, particularly mandarin oranges, holds significant economic importance in Bangladesh.
• Accurate identification of mandarin varieties is crucial for effective cultivation and management.

Purpose of the Study:

• To develop and present a deep learning-based approach for classifying mandarin leaves.
• To create an open-source dataset of healthy mandarin leaves for agricultural research.

Main Methods:

• Collected 1917 high-resolution images (2608 × 4624 pixels) of four mandarin varieties in Bangladesh.
• Augmented the dataset with 8000 images, incorporating rotations and contrast enhancement.
• Utilized a deep learning model for image classification of healthy mandarin leaves.

Main Results:

• The deep learning approach achieved reasonable accuracy in classifying mandarin leaf varieties.
• The dataset, comprising healthy leaf images, facilitates accurate classification and early detection of healthy foliage.
• The open dataset supports researchers in advancing leaf classification techniques in agriculture.

Conclusions:

• Deep learning offers a viable solution for accurate mandarin leaf classification.
• The provided dataset is a valuable resource for agricultural research and the development of automated systems.
• This work contributes to enhancing agricultural practices through advanced image classification techniques.