ECZEMA CLASSIFICATION USING MODIFIED CONVOLUTIONAL NEURAL NETWORK RESNET – 50

Abstract

Skin disorders are serious health issues that impact people of all ages, and many of them still lack a proven treatment. The diagnosis and classification of skin diseases could be greatly enhanced by recent developments in artificial intelligence (AI), especially through deep learning techniques. Early diagnosis made possible by these methods can improve patient outcomes and survival rates. Deep learning models are more effective and possibly more accurate than traditional machine learning techniques since they require less operator involvement for feature extraction. Common inflammatory skin conditions like eczema are typified by dry, itchy, and irritated skin areas. Although the precise etiology of eczema is still unknown, a mix of environmental and genetic factors are thought to be responsible. Convolutional Neural Networks (CNNs) are a specific kind of artificial neural network made to interpret structured grid data, like pictures. It is an essential part of contemporary artificial intelligence (AI) and has transformed a number of domains, including computer vision, by making precise and effective picture processing possible. This study compares many pre-trained deep learning models in terms of improved performance and offers a suggested model method for the classification of eczema conditions. A specially created dataset of eczema photos with two distinct classes—each of which shows how the condition varies from the other—was used to train the machine. Using the suggested methodology, this study successfully classified two forms of eczema with a 97% accuracy rate. In addition to dermatologists and primary care professionals, scientists in the relevant field can utilize this method to reliably classify eczema.