Image Captioning on NoCaps (CIDEr-D, CLIP-S, Ref-S)
42.1CIDEr-D (In)LiMBeR + AAPE
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| LiMBeR + AAPEEvaluation Protocol=train/finetune2024.10 | 42.1 | 49.8 | 44.2 | 47.3 | 77.6 | 81.7 | |
| LiMBeR + AAPEEvaluation Protocol=zero-shot2024.10 | 36.1 | 48.8 | 42.9 | 45.1 | 76.3 | 80.3 | |
| LiMBeR2024.10 | 34.3 | 48.4 | 41.6 | 43.9 | 74.7 | 79.4 | |
| MAGMA2024.10 | 30.4 | 43.4 | 36.7 | 38.7 | 74.3 | 78.7 |