Comparison Of Deep Learning Models for Monkeypox Disease Detection

The Monkeypox virus is a species of double stranded DNA virus that causes Monkeypox disease in humans and other mammals. The emergence of the Monkeypox virus poses a fresh threat to global health at a time, when the world continues to try to heal from the damage brought in by the widespread of COVID-19. Although the Monkeypox diseases are not that harmful or contagious as COVID-19, many new cases are reported from various nations across the world. As a result, it won’t be shocking if there is another pandemic, because of insufficient precautions taken. Due to its resemblance to both chickenpox and measles, Monkeypox can be challenging to clinically diagnose in its initial stages. Computer aided Monkeypox lesion detection may be useful for surveillance and the quick identification of suspected cases in circumstances where there is a difficulty in performing confirmatory Polymerase Chain Reaction (PCR) tests. Deep learning (DL) methods has shown enormous promise for image-based diagnosis including lung cancer detection, skin diseases detection, COVID-19 etc. Hence, the same application may be used to detect the Monkeypox diseases too. This paper evaluates four pre trained deep learning models (Squeeznet, DenseNet201, EfficientNetB0, ResNet50v2) for Monkeypox disease detection and six established metrics (Precision, Recall, Sensitivity, Specificity, F1-score and Accuracy) are used to analyze the results. This paper helps to identify the best pretrained model for Monkeypox diseases detection. A publicly available dataset is used in this work.