Prediction and Classification of Skin Diseases Using Convolution Neural Network Techniques

This research paper proposes the dermatological diseases are very predominant diseases around the world. Even though being general, treatment is very problematic and needs broad knowledge in subject. The skin illnesses are mainly caused by bacteria, fungal infection, viruses, allergy, etc. The Photonics-based medical technology and lasers advancement will be utilized for the treatment of skin illnesses which produce results rapidly and precisely. But the medical equipments for such diagnosis are limited and very costly. In such cases deep learning methods found to be helpful. The skin illness detection at an initial phase is a significant part in treatment. Deep learning technique like Convolution Neural Network (CNN) may help to find the problem at an initial stage. Computer vision and deep learning are dual stages which we used to identify diseases accurately. The utilization of deep learning methods has decreased the requirement for human supervision on a regular basis. A dataset of 5,633 images which are divided into five categories have been taken for skin diseases classification. They comprise acne, eczema, melanoma, psoriasis, and urticaria hives. The model also provides the precautions needed to be taken and some recommended medicines for the skin disease. By utilizing CNN algorithm, 83% accuracy is achieved in classification of skin disease.