Arc2Face: A Foundation Model for ID-Consistent Human Faces

TL;DR

Arc2Face is a novel foundation model that generates highly ID-consistent, photo-realistic human face images from ArcFace embeddings, leveraging an upsampled large-scale dataset and adapting a pretrained diffusion model.

Contribution

The paper introduces Arc2Face, a face generation model conditioned solely on identity features, improving face similarity and ID consistency over existing methods.

Findings

Achieves higher face similarity than previous models.

Uses an upsampled WebFace42M dataset for better identity reconstruction.

Training on synthetic images improves face recognition performance.

Abstract

This paper presents Arc2Face, an identity-conditioned face foundation model, which, given the ArcFace embedding of a person, can generate diverse photo-realistic images with an unparalleled degree of face similarity than existing models. Despite previous attempts to decode face recognition features into detailed images, we find that common high-resolution datasets (e.g. FFHQ) lack sufficient identities to reconstruct any subject. To that end, we meticulously upsample a significant portion of the WebFace42M database, the largest public dataset for face recognition (FR). Arc2Face builds upon a pretrained Stable Diffusion model, yet adapts it to the task of ID-to-face generation, conditioned solely on ID vectors. Deviating from recent works that combine ID with text embeddings for zero-shot personalization of text-to-image models, we emphasize on the compactness of FR features, which can fully capture the essence of the human face, as opposed to hand-crafted prompts. Crucially, text-augmented models struggle to decouple identity and text, usually necessitating some description of the given face to achieve satisfactory similarity. Arc2Face, however, only needs the discriminative features of ArcFace to guide the generation, offering a robust prior for a plethora of tasks where ID consistency is of paramount importance. As an example, we train a FR model on synthetic images from our model and achieve superior performance to existing synthetic datasets.