Trace Captioning on Flickr30k
20.3CIDErT2D + Noise + Filtered knowledge
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| T2D + Noise + Filtered knowledge2025.10 | 20.3 | 77.3 | |
| T2D + Noise2025.10 | 19.3 | 75.6 | |
| T2D + Mem.2025.10 | 18.8 | 77 | |
| T2D + Diffusion2025.10 | 18.7 | 75 | |
| T2D + Noise + External knowledge2025.10 | 18.5 | 76.2 | |
| MeaCapreproduced=true2025.10 | 12.6 | 69.8 | |
| ViECapreproduced=true2025.10 | 12 | 68.8 | |
| AlphaCLIP2025.10 | 11.8 | 71 | |
| DeCapreproduced=true2025.10 | 11.2 | 71 | |
| Diffusion Bridgereproduced=true2025.10 | 10.7 | 69.8 | |
| IFCapreproduced=true2025.10 | 10 | 67.4 |