Monkeypox Diagnosis with Convolutional Neural Networks Combined with Colour Space Models

Abstract

The rapidly rising incidence of monkeypox has gradually become an emerging infectious disease threat. An effective and quick strategy for early diagnosis potentially can reduce its infectiousness. This paper explores the classification of monkeypox and other skin diseases using CNN models. Specifically, we first increase the amount of data by applying the data augmentation method to the images, and colour feature selection is performed to use the most effective RGB as the image input feature. The presented CNN model is constructed to be applicable to the classification of monkey pox. We evaluate our model's efficacy by contrasting it with the widely accepted transfer learning approaches. Our model achieves an accuracy of 91%, presenting 8-17% higher accuracy than the transfer learning models. The experimental results demonstrate that our CNN model is superior for monkeypox classification. Our model can be a reliable and effective tool for the initial screening of monkeypox.