Machine Learning Enabled Model for Different Skin Disease Prediction

The skin is a remarkable organ in the human body. It frequently suffered from both well-known and uncommon ailments. As a result, diagnosing human skin disorders is the most complex and difficult area of study. The vast majority of instances are not recorded, as there is little infrastructure and support for the medical field. This paper seeks to deal with this issue. The creation of robust classification frameworks for a variety of skin conditions can benefit from the use of machine learning techniques. Out of a few machine learning techniques, we have worked on convolutional neural networks and EfficientNet to offer an effective compound scaling strategy for expanding the model size to improve accuracy. The number of skin disorders pictures that were detected from the skin illness dataset, which allowed for the identification of skin diseases, was shown in the findings as evidence of the success of the recommended (CNN-EfficientNet) system. The system's backbone was EfficientNet, a contemporary, condensed, and exceptionally accurate network design that achieved accuracy of 95%. Our method produces superior results, and this study also reveals the sought-after augmentation policy implemented, which necessitates significant financial investment. Out of all the contenders, it has been discovered that CNN offers the best training precision for the accurate prediction of skin problems. Keywords: convolutional neural networks, efficientNet, skin disease