BD Guava Fruit and Leaf Disease Dataset

Description

Bangladesh is a country located at the northern edge of the tropics. Geographically, it is bordered by the Himalayan mountain range to the north and the Bay of Bengal to the south, both of which strongly influence its climatic conditions. The country experiences a tropical monsoon climate, in which the monsoon system is the dominant component of the annual weather pattern. Bangladesh’s tropical monsoon climate, characterized by high humidity, frequent rainfall, and persistent cloud cover, creates favorable conditions for the development and spread of guava diseases such as algal leaf spot, red rust, scab, and black mold. To support automated disease detection under realistic agricultural conditions, a guava disease dataset was collected during the monsoon season (June–September), with an average temperature of 27.71°C and relative humidity of 85.39%. Recent advances in deep learning, particularly YOLO-based architectures, have demonstrated strong performance in agricultural disease detection, motivating their application in this work. In Version 5 of the dataset, label-level data and metadata were further refined to improve the dataset’s quality and usability. Each image is associated with structured metadata, including class labels, environmental attributes, and image-level information, ensuring improved traceability and interpretability. The dataset follows a standardized hierarchical organization using YOLO-format annotation files (.txt), with a structure such as class_name/images/class_name01.jpg and class_name/labels/class_name01.txt. Each annotation file stores bounding box information in the YOLO format: class_id center_x center_y width height. For example, 0 0.52 0.48 0.30 0.40, where 0 represents the class ID (e.g., algal leaf spot), 0.52 and 0.48 represent the normalized center coordinates of the bounding box, and 0.30 and 0.40 represent the normalized width and height, respectively. This structured format ensures strict class-wise organization, maintains data consistency across all categories, and guarantees compatibility with deep learning frameworks by enabling efficient data loading, preprocessing, and integration into training pipelines. The augmented images and augmentation_config.py file were provided in Version 4 of the dataset (DOI: 10.17632/rgp9xy5gy2.4). The associated source code is also available through the GitHub repository: GitHub Repository: https://github.com/shuvobasak4004/Guava-Multi-Model-Training-Code

Steps to reproduce

• Collect guava leaf and fruit images from field conditions during different environmental settings (preferably monsoon season: June–September in Bangladesh). • Organize the dataset into class-wise folders based on disease categories. • Convert annotations into YOLO format by creating .txt files for each image, where each file contains class_id center_x center_y width height in normalized form. • Structure the dataset in a hierarchical format such as class_name/images and class_name/labels. • Attach metadata to each image, including class label, environmental attributes (temperature, humidity, weather condition), and image-level information. • Validate dataset integrity by ensuring all images have corresponding label files and correct class mappings. • Use Roboflow for dataset annotation to prepare and manage labeled data efficiently for training. • Use the prepared dataset for training YOLO-based deep learning models for guava disease detection. • Deploy the trained model using user-friendly interfaces such as Tkinter or Gradio for real-world application.

Institutions

• Jahangirnagar University

  Dhaka District, Savar

Categories

Licence