Image Captioning Evaluation on Composite
66.2Kendall-c Tau_cHarmonicEval
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| HarmonicEvalReference protocol=Reference-free2024.12 | 66.2 | — | |
| RIGELReference Requirement=Reference-based2026.06 | 66.1 | 61.2 | |
| RefDISCODE-LVFF=true2025.12 | 66 | 61 | |
| DISCODE-LVLVLM=true, Backbone=LLaVA-Next-72B, Metric Category=Reference-free2025.12 | 66 | 61.1 | |
| DISCODEBackbone=LLaVA, Reference Requirement=Reference-free2026.06 | 66 | 61.1 | |
| Ref-HiFiScoreReference-based=true2025.12 | 65.8 | — | |
| HiFi-ScoreMetric Category=Reference-free2025.12 | 65.8 | — | |
| HiFiScoreReference Requirement=Reference-based2026.06 | 65.8 | — | |
| HiFiScoreReference-based=false2025.12 | 65.7 | — | |
| FLEUR-LVLVLM=true, Backbone=LLaVA-Next-72B, Metric Category=Reference-free2025.12 | 65.7 | 60.7 | |
| HiFiScoreReference Requirement=Reference-free2026.06 | 65.7 | — | |
| EXPERTReference-based=false2025.12 | 65 | — | |
| EXPERTReference Requirement=Reference-free2026.06 | 65 | — | |
| RefDISCODE-INFF=true2025.12 | 64.9 | 60.5 | |
| DISCODE-INLVLM=true, Backbone=InternVL-2.5-78B, Metric Category=Reference-free2025.12 | 64.9 | 60.5 | |
| DISCODEBackbone=InternVL, Reference Requirement=Reference-free2026.06 | 64.9 | 60.5 | |
| RIGELReference Requirement=Reference-free2026.06 | 64.3 | 59.5 | |
| Ref-FLEURReference-based=true, Test time [hour]=42025.12 | 64.2 | — | |
| RefFLEURFF=true2025.12 | 64.2 | — | |
| FLEUR-INLVLM=true, Backbone=InternVL-2.5-78B, Metric Category=Reference-free2025.12 | 64.2 | 59.8 | |
| RefFLEURReference protocol=Reference-based2024.12 | 64.2 | — | |
| Ref-FLEURReference Requirement=Reference-based2026.06 | 64.2 | — | |
| FLEURReference-based=false, Test time [hour]=3.72025.12 | 63.5 | — | |
| FLEURLVLM=true, Metric Category=Reference-free2025.12 | 63.5 | — | |
| FLEURReference protocol=Reference-free2024.12 | 63.5 | — | |
| FLEURReference Requirement=Reference-free2026.06 | 63.5 | — | |
| G-VEvalLVLM=true, Backbone=GPT-4o, Metric Category=Reference-free2025.12 | 63 | 58.3 | |
| G-VEvalReference protocol=Reference-free2024.12 | 63 | — | |
| PAC-S++Backbone=ViT-L, Metric Category=Reference-free2025.12 | 62 | — | |
| RefPAC-S++ViT-LFF=false2025.12 | 61.6 | — | |
| RefPAC-S++Backbone=ViT-L, Metric Category=Reference-based2025.12 | 61.6 | — | |
| BLIP2ScoreReference-based=false, Test time [hour]=0.022025.12 | 61.5 | 56.9 | |
| GPT-FLEURReference protocol=Reference-free2024.12 | 61.5 | — | |
| BLIP2-ScoreReference Requirement=Reference-free2026.06 | 61.5 | 56.9 | |
| CLAIRReference-based=true, Test time [hour]=8.32025.12 | 61 | — | |
| CLAIRFF=true2025.12 | 61 | — | |
| CLAIRReference Requirement=Reference-based2026.06 | 61 | — | |
| PearlReference-based=true, Backbone=ViT-B/32, Test time [hour]=0.0432025.12 | 60.4 | 55.8 | |
| PearlReference-based=true, Backbone=ViT-B/32, Test time [hour]=0.0432025.12 | 60.4 | 55.8 | |
| PearlReference Requirement=Reference-based2026.06 | 60.4 | 55.8 | |
| InfoMetIC+Type=w/o ref2023.05 | 59.3 | — | |
| InfoMetIC+Metric Category=Reference-free2025.12 | 59.3 | — | |
| InfoMetIC+Reference protocol=Reference-free2024.12 | 59.3 | — | |
| InfoMetICType=w/o ref2023.05 | 59.2 | — | |
| InfoMetICReference protocol=Reference-free2024.12 | 59.2 | — | |
| ITIScoreZero-shot=true2026.04 | 59.2 | 55.2 | |
| RefPAC-S++Reference-based=true, Backbone=ViT-B/32, Test time [hour]=0.0232025.12 | 59.1 | 54.7 | |
| RefPAC-S++ViT-B2025.12 | 59.1 | 54.7 | |
| RefPAC-S++Reference Requirement=Reference-based2026.06 | 59.1 | 54.7 | |
| Ref-HICEScoreReference-based=true2025.12 | 58.7 | 53.9 | |
| Ref-HICEScoreReference Requirement=Reference-based2026.06 | 58.7 | 53.9 | |
| PearlReference-based=false, Backbone=ViT-B/32, Test time [hour]=0.0432025.12 | 58.4 | 54 | |
| PearlReference-based=false, Backbone=ViT-B/32, Test time [hour]=0.0432025.12 | 58.4 | 54 | |
| PearlReference Requirement=Reference-free2026.06 | 58.4 | 54 | |
| PAC-S++Reference-based=false, Backbone=ViT-B/32, Test time [hour]=0.0132025.12 | 58.3 | 53.9 | |
| PAC-S++Reference Requirement=Reference-free2026.06 | 58.3 | 53.9 | |
| DENEB (ViT-L)FF=false2025.12 | 58.2 | — | |
| DENEBBackbone=ViT-L, Metric Category=Reference-based2025.12 | 58.2 | — | |
| DENEBReference-based=true, Backbone=ViT-B/32, Test time [hour]=0.0382025.12 | 57.9 | 54 | |
| HICEScoreReference-based=false2025.12 | 57.9 | 53.1 | |
| HICE-SMetric Category=Reference-free2025.12 | 57.9 | 53.1 | |
| DENEBReference Requirement=Reference-based2026.06 | 57.9 | 54 | |
| HICEScoreReference Requirement=Reference-free2026.06 | 57.9 | 53.1 | |
| FAIEr-4Type=w/ ref2023.05 | 57.7 | — | |
| FAIEr-4Reference protocol=Reference-based2024.12 | 57.7 | — | |
| PolosReference-based=true, Test time [hour]=0.0362025.12 | 57.6 | 53.7 | |
| PolosFF=false2025.12 | 57.6 | — | |
| PolosMetric Category=Reference-based2025.12 | 57.6 | — | |
| PolosReference protocol=Reference-based2024.12 | 57.6 | — | |
| PolosReference Requirement=Reference-based2026.06 | 57.6 | 53.7 | |
| CAMScoreMetric Category=Reference-free2025.12 | 57.5 | 53.4 | |
| CLIP-StuneType=w/o ref2023.05 | 57.3 | — | |
| RefPAC-SReference-based=true, Backbone=ViT-B/32, Test time [hour]=0.0232025.12 | 57.3 | 53 | |
| RefPAC-S2025.12 | 57.3 | 53 | |
| RefPAC-SReference Requirement=Reference-based2026.06 | 57.3 | 53 | |
| BRIDGEReference-based=false2025.12 | 57.2 | 52.9 | |
| BRIDGEBackbone=ViT-L, Metric Category=Reference-free2025.12 | 57.2 | 52.9 | |
| BRIDGEReference Requirement=Reference-free2026.06 | 57.2 | 52.9 | |
| UMICType=w/o ref2023.05 | 56.1 | — | |
| UMICMetric Category=Reference-free2025.12 | 56.1 | — | |
| UMICReference protocol=Reference-free2024.12 | 56.1 | — | |
| UMICZero-shot=true2026.04 | 56.1 | 55.3 | |
| PAC-SReference-based=false, Backbone=ViT-B/32, Test time [hour]=0.0132025.12 | 55.7 | 51.5 | |
| MIDZero-shot=true2026.04 | 55.7 | 53.9 | |
| PAC-SZero-shot=true2026.04 | 55.7 | 51.5 | |
| PAC-SReference Requirement=Reference-free2026.06 | 55.7 | 51.5 | |
| RefCLIP-SType=w/ ref2023.05 | 55.4 | — | |
| RefCLIP-SReference-based=true, Backbone=ViT-B/32, Test time [hour]=0.0142025.12 | 55.4 | 51.2 | |
| RefCLIP-SFF=true2025.12 | 55.4 | 51.2 | |
| RefCLIP-SMetric Category=Reference-based2025.12 | 55.4 | 51.2 | |
| RefCLIP-ScoreReference protocol=Reference-based2024.12 | 55.4 | — | |
| RefCLIP-SReference Requirement=Reference-based2026.06 | 55.4 | 51.2 | |
| CLIP-SType=w/o ref2023.05 | 53.8 | — | |
| CLIP-SReference-based=false, Test time [hour]=< 0.012025.12 | 53.8 | 49.8 | |
| CLIP-SMetric Category=Reference-free2025.12 | 53.8 | 49.8 | |
| CLIP-SReference protocol=Reference-free2024.12 | 53.8 | — | |
| CLIP-SZero-shot=true2026.04 | 53.8 | 49.8 | |
| CLIP-SReference Requirement=Reference-free2026.06 | 53.8 | 49.8 | |
| ViLBERTScore-FType=w/ ref2023.05 | 52.4 | — | |
| ViLBERTScoreFF=true2025.12 | 52.4 | — |