AI Paper Atlas

Research map / Image generation / Editing and style transfer

Arc2Face: A Foundation Model for ID-Consistent Human Faces

Foivos Paraperas Papantoniou, Alexandros Lattas, Stylianos Moschoglou, Jiankang Deng, Bernhard Kainz, Stefanos Zafeiriou

ECCV 2024 · Image generation › Editing and style transfer · approach: Diffusion face generation and editing · 106 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
train ID-to-face model on WebFace42M instead of FFHQID-conditioned latent diffusion trained from scratch · Synth-500 and AgeDBID similarityImproved
train on 10% of WebFace42M (about 4M images) instead of fullID-conditioned latent diffusion trained from scratch · Synth-500 and AgeDBID similarityNo clear effect
project ArcFace through the fine-tuned CLIP text encoder instead of a 4-layer MLPArc2Face · Synth-500 and AgeDBID similarityImproved
condition only on ID embeddings instead of ID plus text or CLIP image featuresArc2Face · Synth-500 (prompt 'photo of a person')LPIPS diversity: 0.410 (Photomaker) -> 0.506Improved
use ID embeddings without landmark conditioning instead of InstantID's landmarksArc2Face · Synth-500pose l2 diversity: 0.059 -> 0.317Improved
train FR model on Arc2Face synthetic data instead of DCFace data (0.5M)IR-SE-50 + AdaFace · 5-dataset averageverification accuracy: 89.56 -> 91.73Improved
scale synthetic FR training data from 0.5M to 1.2M imagesArc2Face · 5-dataset averageverification accuracy: 91.73 -> 93.14Improved
compare synthetic Arc2Face data with real CASIA-WebFace for FR trainingIR-SE-50 + AdaFace · 5-dataset averageverification accuracy: 94.62 (real) vs 91.73Hurt
reduce ArcFace embeddings to fewer PCA componentsArc2Face · WebFace42M ID embeddingsfacial fidelityHurt
average ArcFace embeddings of multiple images of same personArc2Face · 400 AgeDB IDsID similarityImproved
check memorization via closest training imageArc2Face · Synth-500 unseen IDsArcFace similarity: 0.37 to closest train image vs 0.74 to inputNo clear effect

Later papers that compared with this method

8 later accepted papers report Arc2Face (10K) as a baseline in their results tables; 7 of them report a better number on most of the results they share with it.

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy