The SREFV project aims at extending prior research on synthesis of realistic faces, to support the generation of videos containing animated faces with synthetic identities, depicted either frontally or in varying poses.
Besides the obvious artistic and entertainment purposes, the outcome of this project will constitute an interesting tool to de-identify and diversify the faces depicted in video training datasets, helping to protect the identity of volunteers, and to mitigate eventual age, gender, and ethnic dataset collection biases.
- Prof. Patrick Flynn (PI)
- Prof. Kevin Bowyer (PI)
- Prof. Adam Czajka (PI)
- Prof. Walter Scheirer (PI)
- Daniel Moreira (Postdoc)
- Sandipan Banerjee (PhD Student)