Generating fake human faces using One Class SVM
Mohamed El Boujnouni, Mohamed Jedra · 2022 2nd International Conference on Innovative Research in Applied Science, Engineering and Technology (IRASET) · 2022
This work presents a new model that generates automatically new fake human faces after learning from a training set containing real ones. The proposed model, based on One Class Support Vector Machine (OCSVM), was formulated mathematically and an effective iterative algorithm was proposed to implement it. Several tests, using a large dataset of human face pictures, have shown that our method can generate successfully face images of persons that don't really exist.