Image Retrieval on Flickr 2.4M (test)
4.359Ave AesGPT-4o
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GPT-4oVLM=N/A, Aesthetic Condition=L, Relevance Condition=L, Controllable=false2026.02 | 4.359 | 0.378 | |
| QCQCVLM=CoCa, Aesthetic Condition=L, Relevance Condition=L, Controllable=true2026.02 | 4.458 | 0.355 | |
| QCQCVLM=CoCa, Aesthetic Condition=L, Relevance Condition=H, Controllable=true2026.02 | 4.53 | 0.391 | |
| QCQCVLM=Blip2, Aesthetic Condition=L, Relevance Condition=L, Controllable=true2026.02 | 4.541 | 0.353 | |
| QCQCVLM=CoCa, Aesthetic Condition=L, Relevance Condition=M, Controllable=true2026.02 | 4.615 | 0.366 | |
| LLaMA3VLM=N/A, Aesthetic Condition=L, Relevance Condition=L, Controllable=false2026.02 | 4.73 | 0.351 | |
| PrefixVLM=N/A, Aesthetic Condition=L, Relevance Condition=L, Controllable=false2026.02 | 4.735 | 0.35 | |
| FTVLM=CoCa, Aesthetic Condition=L, Relevance Condition=L, Controllable=false2026.02 | 4.756 | 0.365 | |
| PTVLM=N/A, Aesthetic Condition=L, Relevance Condition=L, Controllable=false2026.02 | 4.776 | 0.345 | |
| FTVLM=Blip2, Aesthetic Condition=L, Relevance Condition=L, Controllable=false2026.02 | 4.795 | 0.37 | |
| QCQCVLM=CoCa, Aesthetic Condition=M, Relevance Condition=H, Controllable=true2026.02 | 4.841 | 0.386 | |
| QCQCVLM=CoCa, Aesthetic Condition=M, Relevance Condition=M, Controllable=true2026.02 | 4.852 | 0.36 | |
| QCQCVLM=Blip2, Aesthetic Condition=M, Relevance Condition=M, Controllable=true2026.02 | 4.906 | 0.366 | |
| QCQCVLM=CoCa, Aesthetic Condition=M, Relevance Condition=L, Controllable=true2026.02 | 4.934 | 0.354 | |
| QCQCVLM=CoCa, Aesthetic Condition=H, Relevance Condition=M, Controllable=true2026.02 | 5.17 | 0.368 | |
| QCQCVLM=Blip2, Aesthetic Condition=H, Relevance Condition=H, Controllable=true2026.02 | 5.191 | 0.39 | |
| QCQCVLM=CoCa, Aesthetic Condition=H, Relevance Condition=L, Controllable=true2026.02 | 5.222 | 0.353 | |
| QCQCVLM=CoCa, Aesthetic Condition=H, Relevance Condition=H, Controllable=true2026.02 | 5.27 | 0.39 |