Reproduce “Image-Based Virtual Try-on System With Clothing-Size Adjustment” 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

Minoru Kuribayashi, Koki Nakai, Nobuo Funabiki

Researchers at Okayama University developed an image-based virtual try-on system capable of adjusting clothing size, a feature lacking in prior systems. Their method integrates real-world person and garment dimensions to dynamically resize clothing within the try-on image, producing visually distinct and size-accurate outputs preferred by users in subjective evaluations.

Key insights

  • Integrating quantifiable real-world size data with pose estimation enables precise control over virtual garment dimensions within image generation pipelines.
  • Selective resizing of only the clothing area in segmentation maps, rather than the entire person, is an effective strategy for dynamic size adjustment in VTON.
  • Post-processing corrections are crucial for maintaining visual quality and natural appearance when manipulating specific garment regions, preventing artifacts like gaps or misalignments.

Summary by alphaXiv.

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