Identification of Melanoma Lesions using Deep Learning Networks and Cloud Computing

An integration of computer vision algorithms, machine learning, and deep learning to extract information from digital photos is particularly important in dermatology for the early identification and treatment of diseases like melanoma. Two very important issues in the field of melanoma detection research are discussed in this article. The first argument highlights the importance of performing regular training and testing cycles, as well as the risks associated with transfer learning when even little changes to dataset parameters might affect classifier performance. To address the second point’s call for a system architecture that can adapt to shifting training datasets, proposed a Cloud-Fog-Edge Computing hybrid architecture for a Melanoma Detection service. The distributed technique considerably shortens the running time of continuous retrain, as shown by studies done on both a single computer and a variety of distribution systems.