Recognition of human skin diseases using inception-V3 with transfer learning

Skin disease is an irritable disease and may be the motive of deadly to human life. So, all of us ought to be aware of this alarming health problem. Recognition of skin diseases is a very challenging task because of its various characteristics. To avoid delay in treatment, in this paper, five most common skin diseases: Vascular lesion, Solar lentigo, Actinic keratosis, Squamous cell carcinoma, and Basal cell carcinoma have been investigated through the Inception-V3 with and without transfer learning. An extensive experiment is performed, and the model’s effectiveness is tested through standard metrics such as accuracy, F1 score, and AUC of the Receiver Operating Characteristics (ROC) curve. Inception-V3 with transfer learning has achieved the highest test accuracy of 98.16%. The obtained results are also compared with the state-of-the-art approaches.