AZINDI: A Multi-Class Dataset for Deep Learning Based Azadirachta Indica Leaf Disease and Health Recognition

Description

Azadirachta Indica Leaf Disease and Health Classification Dataset is a curated image dataset developed for research in artificial intelligence, machine learning, deep learning, computer vision, and automated plant disease classification. The dataset contains leaf images of Azadirachta indica (Neem) collected from Rajbari and Ashulia, Dhaka, Bangladesh. The dataset is organized into three classes: Chlorotic, Disease, and Healthy. To support balanced model training, each class contains 1,000 training images, 150 validation images, and 150 test images, resulting in 3,000 training images, 450 validation images, and 450 test images, with 4,500 images in total. The dataset is intended for developing and evaluating deep learning and computer vision models for Neem leaf health assessment, disease classification, feature extraction, and explainable AI-based agricultural applications. The images can be used for benchmarking CNN, Transformer, and hybrid deep learning architectures.

Steps to reproduce

Dataset Preparation Steps 1. Image Collection 2. Class Categorization 3. Image Cleaning and Background Removing 4. Data Augmentation 5. Image Resizing 6. Class Balancing 7. Train–Validation–Test Splitting 8. Quality Verification 9. Final Dataset Organization

Institutions

• Daffodil International University

  Dhaka Division, Dhaka

Categories

Licence