Reproduce “UniFit: Towards Universal Virtual Try-on with MLLM-Guided Semantic Alignment” 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
Wei Zhang, Yeying Jin, Xin Li, Yan Zhang, Xiaofeng Cong, Cong Wang, Fengcai Qiao, zhichao Lian
UniFit pairs instructions and reference images to guide try-on and garment reconstruction, using self-synthesized examples for six task categories. Its semantic guidance is compared with a T5 replacement on single-garment try-on using VITON-HD: SSIM is 0.887 versus 0.851, and LPIPS is 0.071 versus 0.098. This supports the combined module in that setting, not uniform superiority across tasks.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.