Contextualized Personal Instance Retrieval on DeepFashion2 1.0 (test)
38.4MRRMeta-Personalized Vision-Language Model
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Meta-Personalized Vision-Language ModelBackbone=ViT-B/162023.06 | 38.4 | 51.4 | |
| Meta-Personalized Vision-Language ModelBackbone=ViT-B/322023.06 | 34.4 | 45.2 | |
| PALAVRABackbone=ViT-B/322023.06 | 28.4 | 39.2 | |
| CLIP (language)2023.06 | 21.2 | 25.3 | |
| CLIP (V+L)2023.06 | 20.8 | 25.4 | |
| CLIP (visual)2023.06 | 14.2 | 17.3 | |
| COLLIEBackbone=ViT-B/322023.06 | 7.9 | 0.7 | |
| AdapterBackbone=ViT-B/322023.06 | 5.9 | 0.7 | |
| Random2023.06 | 2.9 | 1.8 |