Reproduce “NGL: Natural Garment Language for Training-Free Sewing Pattern Estimation” 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
Anna Badalyan, Pratheba Selvaraju, Giorgio Becherini, Omid Taheri, Victoria Fernandez Abrevaya, Michael Black
The authors present NGL, a training-free method for turning clothing images into sewing patterns. A frozen vision-language model describes garment layers using 44 semantic attributes, and a deterministic parser maps them to GarmentCode templates, which generate patterns. On 36 Dress4D images, NGL-GPT-5 reports CD 1.75 and F-score 0.81, versus 2.69 and 0.79 for Design2GarmentCode-GPT-5; this subset comparison does not establish that representation alone explains the difference.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.