Image-Text Retrieval on MSCOCO 5K
69.78I-T ScoreHarmoCLIP
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| HarmoCLIPBackbone=ViT-B/16, Extra data / VLM augmentation=false2025.11 | 69.78 | — | — | — | |
| HarmoCLIPBackbone=ViT-B/16, Data Enhancement=false2025.11 | 69.78 | — | — | — | |
| Fully-FTArchitecture=VSE∞2023.08 | 64.2 | 51.2 | 468.9 | — | |
| Fully-FTParams. (M)=201.2, Mem. (G)=176.82026.04 | 64.2 | 51.2 | 468.9 | — | |
| Fully-FTParams. (M)=201.2, Mem. (G)=176.82026.04 | 64.2 | 51.2 | 468.9 | — | |
| FG-CLIPBackbone=ViT-B/16, Extra data / VLM augmentation=true2025.11 | 64.1 | — | — | — | |
| FG-CLIPBackbone=ViT-B/16, Data Enhancement=true2025.11 | 64.1 | — | — | — | |
| Ours‡Params. (M)=11.8, Mem. (G)=25.42026.04 | 63.4 | 48.7 | 453.8 | — | |
| Ours♠Params. (M)=11.4, Mem. (G)=25.2, QPS=14.832026.04 | 63.2 | 48 | 452.3 | — | |
| SHERLParams. (M)=11.3, Mem. (G)=24.4, QPS=10.242026.04 | 62.5 | 47.3 | 450.8 | — | |
| SHERLParams. (M)=11.3, Mem. (G)=24.42026.04 | 62.5 | 47.3 | 450.8 | — | |
| Ours†Params. (M)=12.9, Mem. (G)=25.52026.04 | 62.2 | 47.1 | 449.1 | — | |
| Ours♡Params. (M)=12.4, Mem. (G)=25.1, QPS=14.712026.04 | 61.9 | 46.8 | 448.1 | — | |
| UniPTArchitecture=VSE∞2023.08 | 61.1 | 45.9 | 445.3 | — | |
| UniPTParams. (M)=12.4, Mem. (G)=24.4, QPS=8.612026.04 | 61.1 | 45.9 | 445.3 | — | |
| UniPTParams. (M)=12.4, Mem. (G)=24.42026.04 | 61.1 | 45.9 | 445.3 | — | |
| LSTArchitecture=VSE∞2023.08 | 57.8 | 43.1 | 434.5 | — | |
| LSTParams. (M)=9.7, Mem. (G)=24.42026.04 | 57.8 | 43.1 | 434.5 | — | |
| LSTParams. (M)=9.7, Mem. (G)=24.42026.04 | 57.8 | 43.1 | 434.5 | — | |
| EVA-CLIPBackbone=ViT-B/16, Extra data / VLM augmentation=false2025.11 | 57.32 | — | — | — | |
| EVA-CLIPBackbone=ViT-B/16, Data Enhancement=false2025.11 | 57.32 | — | — | — | |
| FineCLIPBackbone=ViT-B/16, Extra data / VLM augmentation=true2025.11 | 54.5 | — | — | — | |
| FineCLIPBackbone=ViT-B/16, Data Enhancement=true2025.11 | 54.5 | — | — | — | |
| CLIPBackbone=ViT-B/16, Extra data / VLM augmentation=false2025.11 | 51.8 | — | — | — | |
| CLIPBackbone=ViT-B/16, Data Enhancement=false2025.11 | 51.8 | — | — | — | |
| Partially⇓Architecture=VSE∞2023.08 | 51.2 | 37.6 | 401.7 | — | |
| CLIPSelfBackbone=ViT-B/16, Extra data / VLM augmentation=false2025.11 | 18.8 | — | — | — | |
| CLIPSelfBackbone=ViT-B/16, Data Enhancement=false2025.11 | 18.8 | — | — | — | |
| RegionCLIPBackbone=ViT-B/16, Extra data / VLM augmentation=false2025.11 | 3.4 | — | — | — | |
| RegionCLIPBackbone=RN50, Data Enhancement=false2025.11 | 3.4 | — | — | — | |
| CHANVisual Backbone=Faster R-CNN, Text Encoder=BERT, Ensemble=false2026.03 | — | — | 443.9 | — | |
| CLIPBackbone=ViT-B/16, Data Enhancement=false2025.11 | — | — | — | 37.6 | |
| CLIPSelfBackbone=ViT-B/16, Data Enhancement=false2025.11 | — | — | — | 16.1 | |
| CMSFVisual Backbone=Fast R-CNN, Text Encoder=BiGRU, Ensemble=false2026.03 | — | — | 432 | — | |
| CMSFVisual Backbone=Faster R-CNN, Text Encoder=BERT, Ensemble=false2026.03 | — | — | 445.8 | — | |
| EVA-CLIPBackbone=ViT-B/16, Data Enhancement=false2025.11 | — | — | — | 40.47 | |
| FG-CLIPBackbone=ViT-B/16, Data Enhancement=true2025.11 | — | — | — | 45.4 | |
| FineCLIPBackbone=ViT-B/16, Data Enhancement=true2025.11 | — | — | — | 40.2 | |
| HarmoCLIPBackbone=ViT-B/16, Data Enhancement=false2025.11 | — | — | — | 53.44 | |
| HREMVisual Backbone=Faster R-CNN, Text Encoder=BERT, Ensemble=false2026.03 | — | — | 444.1 | — | |
| MaxMatchVisual Backbone=Faster R-CNN, Text Encoder=BERT, Ensemble=false2026.03 | — | — | 446.5 | — | |
| MMCAVisual Backbone=Faster R-CNN, Text Encoder=BERT, Ensemble=false2026.03 | — | — | 416.4 | — | |
| NAAFVisual Backbone=Fast R-CNN, Text Encoder=BiGRU, Ensemble=false2026.03 | — | — | 430.9 | — | |
| RegionCLIPBackbone=RN50, Data Enhancement=false2025.11 | — | — | — | 2 | |
| SCANVisual Backbone=Fast R-CNN, Text Encoder=BiGRU, Ensemble=true2026.03 | — | — | 410.9 | — | |
| TERANVisual Backbone=Faster R-CNN, Text Encoder=BERT, Ensemble=true2026.03 | — | — | 443.4 | — | |
| USERVisual Backbone=Faster R-CNN, Text Encoder=BERT, Ensemble=false2026.03 | — | — | 445.5 | — | |
| VSE∞Visual Backbone=Fast R-CNN, Text Encoder=BiGRU, Ensemble=false2026.03 | — | — | 421.9 | — | |
| VSE∞Visual Backbone=Faster R-CNN, Text Encoder=BERT, Ensemble=false2026.03 | — | — | 434.2 | — | |
| VSRNVisual Backbone=Fast R-CNN, Text Encoder=BiGRU, Ensemble=true2026.03 | — | — | 415.7 | — | |
| VSRN++Visual Backbone=Faster R-CNN, Text Encoder=BERT, Ensemble=true2026.03 | — | — | 425.4 | — |