Monkeypox Skin Lesion Dataset

Objective

The primary goal of the “Monkeypox Skin Lesion Dataset” is to distinguish monkeypox cases from similar non-monkeypox cases. In addition to the ‘Monkeypox’ class, the dataset includes images of skin lesions from ‘Chickenpox’ and ‘Measles’ under a class name ‘Others’, facilitating binary classification.

Content

The dataset comprises three folders:

1. Original Images: This folder contains 228 images in total, with 102 images belonging to the ‘Monkeypox’ class and the remaining 126 images representing the ‘Others’ class (non-monkeypox cases such as chickenpox and measles).
2. Augmented Images: To enhance the classification task, various data augmentation techniques such as rotation, translation, reflection, shear, hue, saturation, contrast, brightness jitter, noise addition, and scaling have been applied using MATLAB R2020a.
3. Although these augmentations can be perform using ImageGenerator or other image augmentors, we provide the augment images to ensure reproducibility. Post-augmentation, the number of images increase approximately 14-fold, resulting in 1428 images for the ‘Monkeypox’ class and 1764 images for the ‘Others’ class.
4. Fold1: This folder contains one of the three-fold cross-validation datasets. To avoid bias in training, three-fold cross-validation was performed. The original images were split into training, validation, and test sets in an approximate ratio of 70:10:20, maintaining patient independence. Only the training and validation images were augment, while the test set contains only the original images. Users can use the folds directly or employ other algorithms to augment the original data.

Web Application

Given the rapid community transmission of monkeypox, a consumer-level software is necessary to raise awareness and prompt rapid action. We develop a user-friendly web application name Monkey Pox Detector using the open-source Streamlit framework, which utilizes our trained model to predict whether a user should see a specialist. The application provides prediction accuracy and will continue to improve with future updates, incorporating user data. The web app features a Flask core, allowing for cross-platform deployment in the future.