Melanoma Detection Using Two Neural Networks. A Comparison Study

If not detected and treated in time, melanoma is one of the most dangerous skin cancers. This article analyzes two models of neural networks, adapted for the classification of skin lesions in melanoma and nonmelanoma: InceptionV3 and MobileNetV2. Transfer learning methods and the HAM10000 database were used for learning. Finally, from the analysis of statistical and time performance, recommendations were made for computer or mobile applications.