Synthesizing Coupled 3D Face Modalities by TBGAN

 

The Synthesizing Coupled 3D Face Modalities by TBGAN is a 3D GAN model of facial shape, texture, normals and expression built from nearly 3,500 individuals.

 

Acquiring the TBGAN model

We are now accepting applications for access to the TBGAN for research purposes. Applications should be made by signing the following End User License Agreement:

EULA

Applications should be made by an academic member of staff or manager.

 

Acquiring the TBGAN code

The code used to train Trunk-Branch Generative Adversarial Network (TBGAN) is available on Github:

github.com/barisgecer/TBGAN

This repository also contains further technical details on how to use the model to synthesize 3D face data

 

Further information

For more information, please see the publication: Synthesizing Coupled 3D Face Modalities by Trunk-Branch Generative Adversarial Networks, or contact Baris Gecer.