Digital Image Detection of Monkeypox Disease using Convolutional Neural Network Algorithm with MobilenetV2 Architecture

Abstract

In July 2022, monkeypox was declared a global health emergency due to its occurrence in over 70 countries. The first case of monkeypox in Indonesia emerged in Jakarta in August 2022. The challenge faced by healthcare workers in distinguishing between monkeypox, chickenpox, and measles— ailments sharing similar symptoms—prompted the initiation of a research study. This study aimed to develop an automated algorithm for detecting digital images of monkeypoxes. The algorithm used was a convolutional neural network with MobileNetV2 architecture, implementing transfer learning. The model was trained for a total of 5 epochs and utilized two types of optimizers: Adam and RMSprop. Applying the Adam optimizer with a learning rate of resulted in a test accuracy of 94%, training accuracy of 92%, and a loss function value of 27%. Conversely, the implementation of the RMSprop optimizer, with a learning rate of , resulted in a test accuracy of 97%, and a training accuracy of 97%, albeit with a slightly higher loss function value of 52%. The results indicate that the Adam optimizer may be more effective in fine-tuning the model parameters to optimize the detection of monkeypox images during training