|Topic:||Human Face Synthesis|
|Supervisor:||Ing. Vojtěch Franc Ph.D.|
|Description:||The current state-of-the-art methods for face recognition are based on deep
Convolutional Neural Networks which rely on large databases of annotated facial
images. For example, learning predictor of biological age requires large set of
examples of facial images along with the age of captured subjects. Collection of such example sets is expensive for several reasons. The task of this project
will be to develop a method to augment the limited set of annotated real faces
by generating synthetic photo-realistic faces with the same attributes. The
success of the method will be evaluated via measuring accuracy improvement of a
CNN learned from the generated synthetic images.
|Bibliography:||- G.Antipov, M.Baccouche, J.L.Dugelay. Face Aging With Conditional Generative Adversarial Networks. ArXiv 2017.
- S.Bazrafkan, H.Javidnia, P.Corcoran. Face Synthesis with Landmark Points from Generative Adversarial Networks and Inverse Latent Space Mapping. ArXiv 2018.
- Z.Lu, Z.Li, J.Cao, R.He, Z. Sun. Recent Progress of Face Image Synthesis. ArXiv 2017.
|Realization form:||SW projekt|