Fake It Till You Make It: Face Analysis in the Wild Using Synthetic Data Alone

The community has long enjoyed the benefits of synthesizing training data with graphics, but the domain gap between real and synthetic data has remained a problem, especially for human faces. Researchers have tried to bridge this gap with data mixing, domain adaptation, and domain-adversarial training, but we show that it is possible to synthesize data with minimal domain gap, so that models trained on synthetic data generalize to real in-the-wild datasets.

In this presentation, the team describes how to combine a procedurally-generated parametric 3D face model with a comprehensive library of hand-crafted assets to render training images with unprecedented realism and diversity. We train machine learning systems for face-related tasks such as landmark localization and face parsing, showing that synthetic data can both match real data inaccuracy as well as open up new approaches where manual labeling would be impossible.

Image Credit: © Microsoft. All Rights Reserved.

#spark2022 #conference #AI #vfx #vfxindustry #vfxai #artificialintelligence #microsoft #faceanalysis #data #syntheticdata #facemodel

Follow us on Social Media:

Thank you to our sponsors!

TERABYTE: Creative BC.

MEGABYTE: AMPD Technologies · Animal Logic · Capilano University · Netflix · TrickFilm.

KILOBYTE: Atomic Cartoons · Bardel Entertainment Inc. · BRON Media Corp. · DNEG · Image Engine · Kickstart Entertainment · Mainframe Studios · Sony Pictures Imageworks.

BYTE: AMD Canada · Animasia BC · Centre for Digital Media · CMPA BC · Departure Lounge · DigiBC · Foundry · Ghost VFX · NGX Interactive · NOX VFX · Scanline VFX · The Little Dev Shop · Unreal Engine · Vancouver Film School · Versatile Media · Wacom Store Canada.

PARTNERS: ACM SIGGRAPH.

ORGANIZERS: Spark CG Society · Vancouver ACM SIGGRAPH.