NeRF brief captioning on HST
59.2S-BERT ScoreLLaNA-7b
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| LLaNA-7bModality=NeRF2024.06 | 59.2 | 61.66 | 9.47 | 14.94 | 17.06 | |
| BLIP-2 FlanT5-xxlModality=Image (FV), Vantage Point=front-view2024.06 | 57.11 | 59.43 | 8.21 | 18.02 | 12.14 | |
| 3D-LLMModality=Mesh + MV, Vantage Point=multi-view2024.06 | 56.07 | 52.13 | 15.94 | 20.71 | 15.22 | |
| LLaVA-vicuna-13bModality=Image (FV), Vantage Point=front-view2024.06 | 55.62 | 55.56 | 6.56 | 11.81 | 14.52 | |
| LLaVA-vicuna-7bModality=Image (FV), Vantage Point=front-view2024.06 | 54.31 | 56.28 | 10.08 | 14.71 | 14.53 | |
| LLaVA-vicuna-13bModality=Image (MV), Vantage Point=multi-view2024.06 | 54.25 | 55.56 | 9.78 | 14.13 | 14.99 | |
| BLIP-2 FlanT5-xxlModality=Image (BV), Vantage Point=back-view2024.06 | 54.11 | 56.37 | 9.09 | 17.38 | 11.79 | |
| LLaVA-vicuna-7bModality=Image (BV), Vantage Point=back-view2024.06 | 51.75 | 52.29 | 8.13 | 13.96 | 14.18 | |
| LLaVA-vicuna-13bModality=Image (BV), Vantage Point=back-view2024.06 | 50 | 50.79 | 9.39 | 12.76 | 14.46 | |
| PointLLM-7bModality=Point cloud2024.06 | 43.4 | 44.5 | 8.53 | 11.64 | 9.97 | |
| GPT4Point-Opt-2.7BModality=Point cloud2024.06 | 43.15 | 42.22 | 12.02 | 18.73 | 13.69 |