Lychee Leaf Disease Dataset: Advancing Agriculture and Sustainability through Machine Learning-Based Diagnosis
Description
The Lychee Leaf Disease Detection Dataset is a carefully collected dataset aiming to promote research, exploration and application in agriculture. This diagnostic mini-tool enables researchers, agricultural experts, and stakeholders to precisely identify and manage lychee leaf diseases. This dataset facilitates early and accurate detection, allowing farmers to take timely action, optimize crop management practices, and improve agricultural productivity. Also, it provides a reference point for comparisons of existing and new machine learning models to inspire and motivate new methods for disease detection. This dataset comprises 3,768 original lychee leaf images labelled as one of the following condition classes: Anthrax Leaf, Bituminous Leaf, Curl Leaf, Deficiency Leaf, Dry Leaf, Felt Leaf, Fungal Leaf Spot, Healthy Leaf, Leaf Blight, and Leaf Gall. The images were captured during field surveys conducted between June 2024 and December 2024 in Melandaha, Jamalpur, and Gazaria, Munshiganj. Using Redmi Note 11s and iPhone 14 Plus smartphones, the dataset reflects diverse environmental conditions, including variations in lighting and field settings. Expert guidance ensured the collection of high-quality images that authentically represent real-world scenarios, addressing potential challenges such as environmental variability. To enhance its utility for machine learning applications, data augmentation techniques—including flipping, brightening, and rotation—were applied, expanding the dataset to 10,000 augmented images. These augmented samples significantly strengthen the dataset's robustness, facilitating the development of accurate and reliable models for disease detection and classification. Key Features of the Dataset Original Dataset: Number of Images: 3,768 Format: .jpg Augmented Dataset: Number of Images: 10,000 Format: .jpg This dataset offers a foundational resource for agricultural research, serving as a reliable benchmark for machine learning-based disease classification systems. It provides an unparalleled opportunity to enhance precision in lychee leaf disease detection and advance the field of agricultural informatics.