Applications of Deep Learning techniques, for instance recognizing and identifying elements of interest in images and videos, have been rising as Machine Learning continues to evolve. One of such is detection and classification of skin anomalies or exanthems, which has been the focus of this project, to help and quicken the diagnostic process by medical experts.




To achieve this, Convolutional Networks CNNs, mainly ResNet, have been used with satisfying results given several limitations due to an extremely limited database. Other networks that were considered and worked on were Siamese Networks, which have given promising results.




Furthermore, to solve the issue with the database the use of Generative Adversarial Networks GANs with mixed results, due to a tendency to easily overfit Nevertheless, with the implementation of Adaptative Discriminator Augmentation, proposed by the authors of SytleGAN2-ADA, this problem can be mitigated.