Chili Leaf Disease Dataset: Annotated Smartphone Images of Anthracnose, Cercospora Leaf Spot, Leaf Curl Disease, and Healthy Leaves in Bangladesh

Description

Introduction The Chili Leaf Disease Dataset contains 1544 images of chili leaves, captured using smartphone cameras in agricultural fields across Bangladesh. The images are divided into four classes: Anthracnose, Cercospora Leaf Spot, Leaf Curl Disease, and Healthy Leaves. This dataset is designed to assist in the development of machine learning models for automated disease detection in chili plants, supporting agricultural innovation and sustainable farming practices. Dataset Overview o Number of Images: 1544 o Classes: 4 — Anthracnose, Cercospora Leaf Spot, Leaf Curl Disease, Healthy Leaves o Image Sources: Captured using smartphones with varying camera specifications: Redmi 12 (50 MP), Redmi 13 (108 MP), and Tecno Spark 8 Pro (48 MP) o Geographical Region: Bangladesh The dataset includes images with varying lighting, backgrounds, and angles to ensure diversity in field conditions. This allows the dataset to reflect real-world conditions that farmers might encounter when diagnosing. Chili Leaf Diseases Anthracnose: A fungal disease caused by Colletotrichum species, leading to dark, sunken lesions on leaves and fruits. Cercospora Leaf Spot: Caused by Cercospora capsici, this disease results in dark, circular spots on leaves, causing premature leaf drop. Leaf Curl Disease: Caused by viruses like ChiVMV and ToLCV, this disease causes leaves to curl, leading to stunted growth and reduced yield. Healthy Leaves: Includes disease-free chili leaves, serving as a baseline for comparison with diseased leaves. Data Collection The images were captured across various chili fields in Bangladesh using the smartphones listed above. These smartphones were chosen for their accessibility and image quality, reflecting conditions under which farmers typically use smartphones for agricultural tasks. Images were taken from various angles and distances, with different lighting conditions, to simulate real-world scenarios. Use Cases Mobile Applications for Farmers: Develop smartphone apps enabling farmers to take pictures of their plants and receive instant diagnoses on disease presence. Precision Agriculture: Assist farmers by providing early disease detection, reducing pesticide use, and improving crop management. Agricultural Research: Support studies in plant pathology and machine learning for improved disease diagnosis and management systems. Conclusion The dataset is publicly available through Mendeley Data and comes in four different folders class-wise containing the raw JPG images and corresponding CSV metadata files.This dataset provides a valuable resource for developing automated systems that assist farmers in Bangladesh and other regions with disease detection and crop management. By leveraging machine learning, this dataset helps reduce reliance on manual inspection, improves crop health monitoring, and supports more sustainable agricultural practices.

Steps to reproduce

1. Data Collection: Smartphones Used: Images were captured using three smartphones: Redmi 12 (50 MP main camera), Redmi 13 (108 MP main camera), and Tecno Spark 8 Pro (48 MP main camera). Geographical Area: Data was collected from agricultural fields across various regions of Bangladesh. Environmental Conditions: Images were taken under different environmental conditions, with varying lighting, angles, and backgrounds, to simulate real-world conditions that farmers would experience when diagnosing diseases in the field. Data Diversity: The images were captured at different times of the day to ensure the dataset represents diverse lighting conditions (e.g., daylight, overcast, early morning). The images reflect both healthy and diseased chili plants. 2. Image Labeling: After data collection, the images were manually labeled into four classes: Anthracnose: A fungal disease characterized by dark, sunken lesions on leaves and fruits. Cercospora Leaf Spot: A fungal disease marked by dark, circular spots on the leaves, which cause premature leaf drop. Leaf Curl Disease: A viral disease that causes the leaves to curl and deform. Healthy Leaves: Healthy, disease-free chili leaves that serve as a control group for comparison. Each image was assigned a label based on visible symptoms, and the data was organized accordingly. 3. Preprocessing: Cleaning: All images were cleaned to remove unnecessary background clutter or irrelevant objects that could hinder model training. Only the chili leaves with visible symptoms were kept for labeling. Organizing: The dataset was organized into separate folders for each class (Anthracnose, Cercospora Leaf Spot, Leaf Curl Disease, and Healthy Leaves) for easy access and proper training in machine learning models. Format: Images were saved in JPEG/PNG format for compatibility with most machine learning platforms. 4. Dataset Structure: The dataset consists of 1544 images, distributed among the four categories as follows: Anthracnose: 351 images Cercospora Leaf Spot: 371 images Leaf Curl Disease: 383 images Healthy Leaves: 439 images The total number of images is divided to provide a balanced dataset for training and validation purposes. 5. Dataset Availability: The dataset is available for public access and is shared under an open license for non-commercial use. It can be used by researchers, developers, and practitioners in the field of plant disease detection, machine learning, and agricultural research. The dataset is uploaded to Mendeley for easy access and collaboration among the research community. 6. Tools and Software: Data collection was done using smartphones, with no additional advanced imaging tools used. Preprocessing involved basic image editing software for cleaning and organizing the dataset into the appropriate folders. The dataset is ready for use with machine learning frameworks, including TensorFlow and PyTorch, for training disease detection models.

Institutions

• Dhaka International University

  Dhaka Division, Dhaka

Categories

Licence