Dense Captioning on Visual Genome
16.2mAPFlexCap + GRIT Boxes
Evaluation Results
| Method | Links | |
|---|---|---|
| FlexCap + GRIT Boxestraining=fine-tuned2024.03 | 16.2 | |
| GRIT_GITBackbone=ViT-B, Pre-training=GIT (re-pretrained on CC3M and CC12M)2022.12 | 15.52 | |
| GRITtraining=fine-tuned2024.03 | 15.52 | |
| GRIT_MAEBackbone=ViT-B, Pre-training=MAE2022.12 | 15.48 | |
| TDC+ROCSU2022.12 | 11.49 | |
| TDC+ROCSUtraining=fine-tuned2024.03 | 11.49 | |
| CAG-Net2022.12 | 10.51 | |
| CAG-Nettraining=fine-tuned2024.03 | 10.51 | |
| COCG2022.12 | 9.82 | |
| COCGtraining=fine-tuned2024.03 | 9.82 | |
| COCD2022.12 | 9.36 | |
| JIVC2022.12 | 9.31 | |
| JIVCtraining=fine-tuned2024.03 | 9.31 | |
| ImgG2022.12 | 9.25 | |
| FCLN2022.12 | 5.39 | |
| FCLNtraining=fine-tuned2024.03 | 5.39 |