Image Captioning on COCO (test)
154.9CIDErOFA
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| OFAModel Type=Autoregressive Model2023.05 | 154.9 | 44.9 | 32.5 | — | — | — | |
| CoDiModel Type=Diffusion Model2023.05 | 149.9 | 40.2 | 31 | — | — | — | |
| SotA2022.04 | 149.6 | — | — | — | — | — | |
| BLIP2Model Type=Autoregressive Model2023.05 | 145.8 | 43.7 | — | — | — | — | |
| Pretrained FT SOTAFT=true, Shot=444K2022.04 | 143.3 | — | — | — | — | — | |
| Flamingomode=Fine-tuned2022.04 | 138.1 | — | — | — | — | — | |
| SCD-NetModel Type=Diffusion Model2023.05 | 131.6 | 39.4 | 29.2 | — | — | — | |
| OscarModel Type=Autoregressive Model2023.05 | 124.12 | 36.58 | 30.4 | — | — | — | |
| DDCapModel Type=Diffusion Model2023.05 | 117.8 | 35 | 28.2 | — | — | — | |
| FlamingoFT=false, Shot=322022.04 | 113.8 | — | — | — | — | — | |
| FlamingoFew-shot=32 shots2022.04 | 113.8 | — | — | — | — | — | |
| ClipCapModel Type=Autoregressive Model2023.05 | 108.35 | 32.15 | 27.1 | — | — | — | |
| Flamingo-9BFT=false, Shot=322022.04 | 106.3 | — | — | — | — | — | |
| FlamingoFT=false, Shot=42022.04 | 103.2 | — | — | — | — | — | |
| Flamingo-3BFT=false, Shot=322022.04 | 99 | — | — | — | — | — | |
| Flamingo-9BFT=false, Shot=42022.04 | 93.1 | — | — | — | — | — | |
| Saliency and Context Attention2017.06 | 89.8 | 28.4 | 24.8 | 52.1 | — | — | |
| Flamingo-3BFT=false, Shot=42022.04 | 85 | — | — | — | — | — | |
| FlamingoFT=false, Shot=02022.04 | 84.3 | — | — | — | — | — | |
| Tavakoli et al.Saliency Map Model=GBVS2017.06 | 84.1 | 28.7 | 23.5 | 51.2 | — | — | |
| Tavakoli et al.Saliency Map Model=iSEEL2017.06 | 83.6 | 28.3 | 23.5 | 50.8 | — | — | |
| Flamingo-9BFT=false, Shot=02022.04 | 79.4 | — | — | — | — | — | |
| LMCAPmode=few-shot2023.05 | 75.9 | 19.9 | 22 | — | — | — | |
| Flamingo-3BFT=false, Shot=02022.04 | 73 | — | — | — | — | — | |
| Zero/Few shot SOTAFT=false, Shot=42022.04 | 32.2 | — | — | — | — | — | |
| ZeroCapmode=zero-shot, language-specificity=English-specific2023.05 | 14.6 | 2.6 | 11.5 | — | — | — | |
| ConZICmode=zero-shot, language-specificity=English-specific2023.05 | 12.8 | 1.3 | 11.5 | — | — | — | |
| AnyDoorBackbone=BLIP-22025.11 | — | 40.7 | 31.6 | 60.9 | 139.5 | 99.7 | |
| BadNetBackbone=BLIP-22025.11 | — | 40.5 | 31.7 | 60.9 | 138.8 | 94.7 | |
| BlendedBackbone=BLIP-22025.11 | — | 40.9 | 31.6 | 61 | 141.1 | 100 | |
| CleanBackbone=BLIP-22025.11 | — | 42.5 | 31.9 | 61.8 | 144.5 | — | |
| CTPBackbone=BLIP-22025.11 | — | 40.8 | 31.5 | 60.9 | 139.9 | 96.2 | |
| ShadowCastBackbone=BLIP-22025.11 | — | 40.6 | 31.7 | 60.9 | 139.2 | 83.3 | |
| VLOODBackbone=BLIP-22025.11 | — | 39.9 | 30.8 | 60 | 135.8 | 99.4 |