The First Workshop on Trustworthy Face Avatars (TrustFA 2026) will take place in conjunction with the 20th IEEE International Conference on Automatic Face and Gesture Recognition, to be held in Kyoto, Japan, from May 25 to 29, 2026.
As face recognition technologies continue to evolve, 2D and 3D photorealistic face avatars are emerging as a powerful new paradigm. These avatars enable users to create and manipulate digital representations of themselves, opening new possibilities for avatar-mediated interaction, identity presence, and telepresence across domains such as healthcare, education, and entertainment. At the same time, they introduce critical challenges related to trust, security, privacy, and authentication.
The TrustFA workshop aims to advance the state of the art in face avatar generation and authentication, with a strong focus on ensuring seamless and secure interactions in these emerging contexts. The workshop will foster collaboration among researchers and practitioners working on avatar technologies, encouraging new insights and research directions that address key challenges such as real-time avatar generation, accurate expression retargeting, and behavior-based biometric authentication.
Scope and Topics
The workshop welcomes contributions in a broad range of topics, including but not limited to:
- Avatar generation, manipulation, and motion transfer: 2D/3D head avatars, neural rendering (NeRFs, Gaussian splatting), audio-driven animation, controllability, identity and pose disentanglement, speed–quality trade-offs, expression tracking fidelity, identity preservation, and visual realism.
- Avatar authentication: impersonation detection, appearance-agnostic driver verification, avatar template access authorization, behavior-based signatures, passive and active safeguards (e.g. watermarking), robustness under varying codecs, bitrates and lighting conditions, cross-generator generalization, and presentation or replay attacks.
- Privacy, safety, and governance: consent management, rights to likeness, watermarking and provenance, on-device versus cloud processing, demographic fairness, unequal detectability, identity binding evidence, and template protection.
- Real-time tracking and control: expression retargeting, audio-driven facial animation, gaze and head pose stabilization, and temporally coherent rendering under resource constraints.
- Evaluation and datasets: protocols for realism, identity fidelity, latency and computational efficiency, human-perceived presence, synthetic-to-real generalization, and data documentation standards.
- Applications and user studies: telepresence in healthcare and education, accessibility, entertainment tools, and AR/VR integration.
The workshop is co-organized by an international team of researchers, including Rubén Tolosana and Rubén Vera-Rodríguez, a member of ELLIS Unit Madrid, reinforcing the unit’s active role in advancing trustworthy and responsible AI research in face analysis and biometric technologies.
More info: TrustFA