Dense Captioning on VG V1.2
42.8mAPControlCap
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ControlCapGT localization=true2024.01 | 42.8 | — | — | |
| GRITGT localization=true2024.01 | 40.3 | — | — | |
| BLIP2GT localization=true2024.01 | 37.9 | — | — | |
| T2D + Mem.2025.10 | 21.31 | 31.9 | 78.8 | |
| T2D + Noise2025.10 | 20.26 | 26.3 | 77 | |
| T2D + Diffusion2025.10 | 19.01 | 30.7 | 77.1 | |
| T2D + Noise + Filtered knowledge2025.10 | 18.66 | 31.9 | 78.9 | |
| ControlCapGT localization=false2024.01 | 18.5 | — | — | |
| T2D + Noise + External knowledge2025.10 | 18.43 | 30.3 | 77.8 | |
| DeCapreproduced=true, applied to image crops=true2025.10 | 17.75 | 24.6 | 77.8 | |
| GRITGT localization=false2024.01 | 16.4 | — | — | |
| RegionCLIP + Mem.2025.10 | 15.85 | 21.7 | 76.7 | |
| CapDetGT localization=false2024.01 | 15.4 | — | — | |
| MeaCapreproduced=true, applied to image crops=true2025.10 | 15.01 | 28.6 | 75.1 | |
| ViECapreproduced=true, applied to image crops=true2025.10 | 14.9 | 26.4 | 74.3 | |
| Diffusion Bridgereproduced=true, applied to image crops=true2025.10 | 14.9 | 20 | 72.8 | |
| AlphaCLIP2025.10 | 14.63 | 19.1 | 73.9 | |
| IFCapreproduced=true, applied to image crops=true2025.10 | 13.7 | 21.3 | 72.5 | |
| DCMSTRGT localization=false2024.01 | 13.4 | — | — | |
| TDCGT localization=false2024.01 | 11.9 | — | — | |
| COCGGT localization=false2024.01 | 10.4 | — | — | |
| JIVCGT localization=false2024.01 | 10 | — | — | |
| COCDGT localization=false2024.01 | 9.8 | — | — | |
| ImgGGT localization=false2024.01 | 9.7 | — | — | |
| FCLNGT localization=false2024.01 | 5.2 | — | — |