Reproduce “Learning Dense Hand Contact Estimation from Imbalanced Data” 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

Daniel Sungho Jung, Kyoung Mu Lee

HACO estimates dense contact across a 3D hand surface from images. It combines balanced contact sampling, which broadens the hand-level patterns seen during training, with vertex-level class-balanced loss, which increases the influence of locally rare labels. Trained on 655,000 images from 14 datasets, it achieves higher F1 in evaluated comparisons, but reported evaluation excludes fully non-contact hands and self-contact.

Summary by alphaXiv.

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