notesum.ai
Published at December 10FiVA: Fine-grained Visual Attribute Dataset for Text-to-Image Diffusion Models
cs.CV
Released Date: December 10, 2024
Authors: Tong Wu1, Yinghao Xu, Ryan Po, Mengchen Zhang, Guandao Yang, Jiaqi Wang, Ziwei Liu, Dahua Lin, Gordon Wetzstein
Aff.: 1Stanford University

| Metrics | DB-Lora | IP-Adapter | DEADiff | StyleAligned | Ours | |
|---|---|---|---|---|---|---|
| User-Study | Sub-Acc | 0.393 | 0.163 | 0.605 | 0.520 | 0.817 |
| Attr&Sub-Acc | 0.240 | 0.150 | 0.260 | 0.298 | 0.348 | |
| CLIP-Score | in-domain | 0.180 | 0.161 | 0.211 | 0.196 | 0.228 |
| out-domain | 0.177 | 0.135 | 0.205 | 0.189 | 0.229 | |