REVIEW 3 cited by
Attention Mesh: High-fidelity Face Mesh Prediction in Real-time
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
We present Attention Mesh, a lightweight architecture for 3D face mesh prediction that uses attention to semantically meaningful regions. Our neural network is designed for real-time on-device inference and runs at over 50 FPS on a Pixel 2 phone. Our solution enables applications like AR makeup, eye tracking and AR puppeteering that rely on highly accurate landmarks for eye and lips regions. Our main contribution is a unified network architecture that achieves the same accuracy on facial landmarks as a multi-stage cascaded approach, while being 30 percent faster.
Forward citations
Cited by 3 Pith papers
-
JOLT3D: Joint Learning of Talking Heads and 3DMM Parameters with Application to Lip-Sync
JOLT3D jointly trains a 3DMM reconstruction network with a talking head generator, then uses FACS mouth blendshapes from a diffusion model to lip-sync videos while preserving the original chin contour.
-
OPEN: A Benchmark Dataset and Baseline for Older Adult Patient Engagement Recognition in Virtual Rehabilitation Learning Environments
OPEN releases landmark and feature data from 35 hours of older adult virtual rehab sessions with engagement, affect, behavior, and context annotations, plus baselines reaching up to 81% accuracy.
-
Towards High-Fidelity, Identity-Preserving Real-Time Makeup Transfer: Decoupling Style Generation
A decoupled makeup-transfer pipeline with synthetic pseudo-ground-truth training achieves impressive numbers, but the evaluation is circular and code is not released.
Discussion (0). Continue with ORCID to comment.