Monkeypox Virus Detection using Deep Learning Methods

Abstract

The fast spread of the recent monkeypox outbreak has become a public health worry in more than 40 nations outside of Africa. Similarly to chickenpox and measles, a clinical diagnosis of monkeypox in the early stages might be difficult. A computer-assisted method of detecting monkeypox lesions could be helpful for surveillance and early case identification in areas where confirmatory Polymerase Chain Reaction (PCR) assays are not easily accessible. As long as enough data is available for training, deep-learning techniques help automate the detection of skin lesions. First, we refreshed the “Monkeypox Skin Lesion (MSL) Dataset,” which includes photos of monkeypox, other, and normal skin lesions. To enhance the sample size, we enrich the data and set up a 3-fold cross-validation experiment. Following this, multiple pre-trained deep learning models distinguish between monkeypox, normal, and other disorders. These models are ResNet50V2, Xception, and MobileNetV2. An ensemble model consisting of all three is also created. The best overall accuracy is reached by Xception, at 96.19%, followed by ResNet50V2 (93.33%) and the MobileNetV2 model (86.67%). To propose using a typical fine-tuned architecture for different Deep Learning (DL) models for the detection of MonkeyPox virus and compare the results. To improve the accuracy of the existing research MVD-DLM.