ALV-Leaf: A Multi-Class Aloe Vera Leaf Image Dataset for Disease and Health Classification
Description
This dataset contains a curated collection of Aloe Vera leaf images developed for research in plant disease detection, computer vision, and deep learning applications. The images were collected from Rajbari, Dhaka, Bangladesh, between January and February 2026 using a OnePlus Nord CE 4 Lite smartphone camera. During data acquisition, detached leaves were placed on a uniform background to ensure clear visibility of leaf morphology and disease characteristics. The dataset consists of four leaf-condition classes: Disease, Dried, Mature Healthy, and Young Healthy. Each class contains 1,000 augmented training images, 150 test images, and 150 validation images, resulting in 4,000 training images, 600 test images, and 600 validation images. During preprocessing and augmentation, techniques such as rotation, horizontal flipping, vertical flipping, brightness and contrast adjustment, Gaussian noise addition, and image sharpening were applied to increase data diversity and improve the robustness of deep learning models. The dataset is accompanied by a structured metadata file in CSV format containing class labels and image distribution information, supporting efficient dataset management and reproducible research experiments.
Files
Steps to reproduce
1. Collect Aloe Vera leaf images from Rajbari, Dhaka, Bangladesh using a OnePlus Nord CE 4 Lite under uniform background conditions. 2. Categorize images into four classes and balance each class to 1000 images. Apply preprocessing including background removal, resizing (512×512), RGB conversion, and normalization. 3. Perform data augmentation using rotation, flipping, brightness/contrast adjustment, Gaussian noise, and sharpening. 4. Organize images into class-wise folders and generate a metadata CSV file.
Institutions
- Daffodil International UniversityDhaka Division, Dhaka