DiFashion introduces Generative Outfit Recommendation (GOR) to synthesize visually harmonious outfits tailored to individual users. It utilizes exceptional diffusion models for parallel image generation, focusing on high fidelity, compatibility, and personalization. Three key conditions guide the generation process: category prompt, mutual condition, and history condition. Extensive experiments on iFashion and Polyvore-U datasets demonstrate DiFashion's superiority over competitive baselines in both PFITB and GOR tasks.
To Another Language
from source content
arxiv.org
Key Insights Distilled From
by Yiyan Xu,Wen... at arxiv.org 03-01-2024
https://arxiv.org/pdf/2402.17279.pdfDeeper Inquiries