Research map / Image generation / Text-to-image diffusion
Text2Outfit: Controllable Outfit Generation with Multimodal Language Models
Yuanhao Zhai, Yen-Liang Lin, Minxu Peng, Larry S. Davis, Ashwin Chandramouli, Junsong Yuan, David Doermann
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add text modality to image tokens (multimodal feature) (ablation) | Text2Outfit · Polyvore text-to-outfit | Avg P@10: 63.76% -> 65.45% | Improved |
| fine-tune LLM with LoRA instead of frozen (ablation) | Text2Outfit · Polyvore text-to-outfit | Avg P@10: 65.45% -> 84.82% | Improved |
| fine-tune LLM with LoRA instead of frozen (ablation) | Text2Outfit · Polyvore text-to-outfit | CP score: 78.62% -> 86.95% | Improved |
| use Local BCE instead of Global CE for attention mask loss (ablation) | Text2Outfit · Polyvore text-to-outfit | Avg P@10: 87.99 -> 89.95 | Improved |
| use Local BCE instead of Global CE for attention mask loss (ablation) | Text2Outfit · Polyvore text-to-outfit | CP score: 85.63 -> 83.43 | Hurt |
| add compatibility loss (ablation) | Text2Outfit · Polyvore text-to-outfit | CP score: 83.43 -> 84.05 | Improved |
| add compatibility loss (ablation) | Text2Outfit · Polyvore text-to-outfit | Avg P@10: 89.95 -> 90.05 | No clear effect |
| predict outfit composition instead of using all 11 categories (seed text) | Text2Outfit · Polyvore seed-to-outfit | Comp. score (Claude-3): 2.59 -> 4.57 | Improved |
| predict outfit composition instead of using all 11 categories (seed image) | Text2Outfit · Polyvore seed-to-outfit | Comp. score (Claude-3): 2.85 -> 4.32 | Improved |
| use fixed ground-truth mask instead of learnable mask | Text2Outfit · Polyvore text-to-outfit | Avg P@10: 90.05 -> 92.56 | Improved |
| use outfit-level LLM model instead of single-item retrieval with Fashion-CLIP (text to image) | Text2Outfit · Polyvore text-to-outfit | Avg P@10: 78.76 -> 90.05 | Improved |
Closest papers by meaning
- ChatGarment: Garment Estimation, Generation and Editing via Large Language Models (CVPR 2025 · similarity 0.86)
- Easy Regional Contrastive Learning of Expressive Fashion Representations (NeurIPS 2024 · similarity 0.85)
- Multitwine: Multi-Object Compositing with Text and Layout Control (CVPR 2025 · similarity 0.84)
- LLM Blueprint: Enabling Text-to-Image Generation with Complex and Detailed Prompts (ICLR 2024 · similarity 0.84)
- Towards Unified Multi-Modal Personalization: Large Vision-Language Models for Generative Recommendation and Beyond (ICLR 2024 · similarity 0.84)
- TexFit: Text-Driven Fashion Image Editing with Diffusion Models (AAAI 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
