Real Fingerprint Classification Model Using Semi-Supervised Generative Adversarial Network (SGAN)

My name is Divine and I am currently working on a real fingerprint detection and classification model using SGAN. However, I am new to the field of machine learning. The concept of my work is to model a real fingerprint classification model using SGAN technique so that the model is able to efficiently detect and classify real fingerprints as well as classify fake fingerprints from novel materials. Hence, the idea is come up with a one class classifier for real fingerprints. I will also like the work to be done on KAGGLE for easy replicating. In addition, important results such as generators losses, discriminator losses, classification accuracies and how the final one class classifier visually classifies the real and fake fingerprints should be included as the attached image.