Reproduce “A Dense Material Segmentation Dataset for Indoor and Outdoor Scene Parsing” 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
Paul Upchurch, Ransen Niu
DMS provides a dataset for training and evaluating pixel-level material segmentation across whole indoor and outdoor scenes. It contains 44,560 images, 52 labels, and more than 3.2 million segments larger than 100 pixels. In a matched-architecture, 25-class comparison, DMS-trained models reached 0.686 mean class accuracy on DMS-Val versus 0.509 for OpenSurfaces-trained models; this advantage is specific to that evaluation setting.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.