Reproduce “Self-Supervised Pretraining for Label-Efficient Facial Expression Recognition using Smart Glasses” 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
Zhengtao Gao, Francesco Bombassei De Bona, Leonardo Alchieri, Ivana Kiprijanovska, Simon Stankoski, Bojan Sofronievski, Hristijan Gjoreski, Marc Langheinrich and 1 more
The authors use self-supervised pretraining on unlabeled recordings from OMG smart glasses, then fine-tune a Vision Transformer with labeled examples to recognize five posed facial expressions. In person-independent evaluation, pretraining improves macro F1 at every tested label budget. With 5% of labels, it reaches 0.845 ± 0.008, versus 0.839 ± 0.006 for training from scratch with 100%; spontaneous expressions remain untested.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.