Reproduce “Robust Face Alignment Using a Mixture of Invariant Experts” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Oncel Tuzel, Tim K. Marks, Salil Tambe

MIX aligns facial landmarks by first matching an estimated shape to a prototype, then applying softly weighted specialized regressors and a learned penalty against implausible drift. On the 300W combined test set of 689 images with 49 landmarks, it scores 0.7810 versus 0.7613 for CFSS at NAUC₀.₂; higher is better. Its affine invariance is limited to 2D affine transformations.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.