Chilli Leaf Disease Image Dataset for Classification and Early Diagnosis in Agriculture

Description

This dataset contains a total of 8,814 high-resolution images (1000×1000) of chilli leaves collected from agricultural fields located in (Ashulia, Narsingdi, Cumilla, Feni, Noakhali, and Laksham) Bangladesh. The dataset is designed to support research in plant disease classification, computer vision, and deep learning-based agricultural AI. The images are categorized into 6 distinct classes: class and description are given below 1. Bacterial (Leaf lesions caused by bacterial infection) 2. Cercospora (Fungal disease with circular spots) 3. Curl Virus (Virus-infected leaves showing curling symptoms) 4. Healthy Leaf (Fresh and disease-free leaves) 5. Nutrient Deficiency (Yellowing and discoloration from poor nutrient supply) 6. Powdery Mildew (White fungal growth on leaf surfaces) No. of Images: 1. Bacterial (1,629) 2. Cercospora (1,898) 3. Curl Virus (1,590) 4. Healthy Leaf (1,647) 5. Nutrient Deficiency (1,207) 6. Powdery Mildew (843) 📌 Total Images: 8,814 📌 Format: JPG and PNG 📌 Resolution: 1000 × 1000 pixels 📌 Color space: RGB 📌 Capture Device: Smartphone 📌 Environment: Real farm conditions — variable lighting, angles & white backgrounds Applications: Image Classification Disease Detection & Monitoring Transfer Learning Deep Learning research in Agriculture Dataset Benchmarking for Vision Models Folder Structure is given below: Chilli_Leaf_Dataset/ ├── Bacterial_Spot/ ├── Cercospora_Leaf_Spot/ ├── Curl_Virus/ ├── Healthy_Leaf/ ├── Nutrient_Deficiency/ └── Powdery_Mildew/

Steps to reproduce

Researchers can download the dataset and use common deep learning frameworks such as TensorFlow or PyTorch for training classification models. A standard train-test split (e.g., 80/20) is recommended. Pre-processing steps may include image resizing (e.g., 224×224), normalization, and data augmentation to reduce overfitting.

Institutions

• Daffodil International University

Categories

Licence