Monkeypox Disease Detection Using Deep Learning Approach

Abstract

Although humanity is still working to rehabilitate from the harm brought on by the widespread distribution of COVID-19, the Monkeypox disease now poses a fresh threat of spreading worldwide. Due to the recent Monkeypox outbreak's unprecedented advancement in more than 111 countries, public health is now at risk. Monkeypox can be difficult to diagnose clinically in its early stages since it resembles both chickenpox and measles. Under the condition that there are enough training examples available, deep learning techniques have been demonstrated to be useful in the accurate identification of skin lesions. Unfortunately, there are currently no comparable datasets for the Monkeypox disease. So we have found the solution to the problem by utilizing transfer learning approaches. We used thoroughly developed other skin disease datasets. The majority of the images are from news websites, blogs, and other media, and some are available as case studies to the public. We did not use data augmentation to increase our dataset. We only used raw data. Several deep learning models that have already been trained are used in the following stage. For our research, we have worked with VGG16, InceptionV3, ResNet50, InceptionResNetV2, and MobileNetV2. Among them, InceptionV3 model has acquired the best accuracy. It has scored 94.56 %. Although the preliminary findings on this particular dataset are encouraging, a greater dataset with a more mixed population is needed to further improve the universal applicability of such models.