Retrieval-Augmented Gaussian Avatars: Improving Expression Generalization

TL;DR

This paper introduces RAF, a retrieval-augmented training method for template-free head avatars, which enhances expression generalization and robustness by leveraging a large expression bank during training.

Contribution

The paper proposes a novel retrieval-augmented training approach that improves expression coverage and identity-expression decoupling without extra annotations or architectural changes.

Findings

RAF improves expression fidelity on the NeRSemble benchmark.

Retrieval augmentation increases expression diversity and robustness.

User study confirms retrieved expressions are perceptually similar.

Abstract

Template-free animatable head avatars can achieve high visual fidelity by learning expression-dependent facial deformation directly from a subject's capture, avoiding parametric face templates and hand-designed blendshape spaces. However, since learned deformation is supervised only by the expressions observed for a single identity, these models suffer from limited expression coverage and often struggle when driven by motions that deviate from the training distribution. We introduce RAF (Retrieval-Augmented Faces), a simple training-time augmentation designed for template-free head avatars that learn deformation from data. RAF constructs a large unlabeled expression bank and, during training, replaces a subset of the subject's expression features with nearest-neighbor expressions retrieved from this bank while still reconstructing the subject's original frames. This exposes the deformation field to a broader range of expression conditions, encouraging stronger identity-expression decoupling and improving robustness to expression distribution shift without requiring paired cross-identity data, additional annotations, or architectural changes. We further analyze how retrieval augmentation increases expression diversity and validate retrieval quality with a user study showing that retrieved neighbors are perceptually closer in expression and pose. Experiments on the NeRSemble benchmark demonstrate that RAF consistently improves expression fidelity over the baseline, in both self-driving and cross-driving scenarios.