BDLitchi: A Field-Collected Bangladeshi Litchi Leaf Disease Dataset for Deep Learning-Based Detection and Classification

This dataset comprises high-resolution images of lychee (Litchi chinensis) plant leaves, collected from agricultural fields in Chor Kodimpara, Ishwardi, Pabna and Ashulia, Savar, Dhaka, Bangladesh. Lychee is an economically important fruit crop in South Asia, especially in Bangladesh, where maintaining plant health is essential for sustainable agriculture, disease prevention, and maximizing crop yields. Early identification and management of leaf diseases are critical for reducing losses and promoting precision farming practices.

The primary aim of this dataset is to facilitate the development and evaluation of machine learning and deep learning models for the detection, classification, and monitoring of litchi leaf diseases. The images were captured under natural conditions and reflect real-world disease symptoms, making the dataset ideal for practical and scalable applications in agriculture technology.

The dataset is organized into 11 distinct classes, covering both healthy and diseased leaves:

Black Spot: 1,045 images

Burned Leaf: 1,171 images

Dried Leaf: 934 images

Fungal Stripe Damage: 922 images

Healthy Leaf: 988 images

Insect Chewing Damage: 1,160 images

Leaf Blight Disease: 965 images

Pest-Affected Dry Leaf: 879 images

Red Rust Disease: 1,079 images

White Spot: 877 images

Yellow Mosaic Virus: 1,074 images

Total Images: 11,094

Image Specifications:

Original Resolution: 3072 × 1080pixels

Compressed Resolution (if needed): 460 × 638pixels

Image Format: JPG

The labeling and disease class validation of this dataset were reviewed and verified by a Sub-Assistant Agriculture Officer from the Department of Agricultural Extension (DAE), Bangladesh, ensuring high reliability and relevance for agricultural research.

This dataset is a valuable asset for researchers in plant pathology, computer vision, precision agriculture, and AI-based crop monitoring. Its diverse classification, real-world field conditions, and expert validation make it suitable for tasks such as image classification, disease detection, segmentation, and the development of intelligent agricultural systems using machine learning techniques.