Reproduce “An analysis of vision-language models for fabric retrieval” 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
Francesco Giuliari, Asif Khan Pattan, Mohamed Lamine Mekhalfi, Fabio Poiesi
The authors evaluate zero-shot text-to-image retrieval for fabric swatches using descriptions generated from images by ChatGPT-4o-mini. They compare freeform descriptions with templates specifying fabric attributes across three pretrained vision-language models and four FID categories. Attribute-based descriptions improve retrieval across all models, with the Perception Encoder strongest, especially for lattice and printed fabrics; even the best results leave fine-grained fabric retrieval challenging.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.