Narrative Action Evaluation on MTL-NAE re-annotated (test)
38.3mAPPrompt-Guided Multimodal Interaction
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Prompt-Guided Multimodal InteractionBackbone=Video Swin Transformer, Vision-Language Model=CLIP2024.04 | 38.3 | 0.943 | 0.34 | 42.2 | 28.2 | 20.5 | 97 | |
| SwinBert2024.04 | 26.1 | 0.881 | 0.706 | 40.2 | 26.4 | 16.2 | 92 | |
| VLTinT2024.04 | 21.4 | 0.868 | 0.82 | 22.5 | 19.9 | 14.4 | 90 | |
| VLCap2024.04 | 19.7 | 0.851 | 0.867 | 19.8 | 18.7 | 13.9 | 90 | |
| UniVL2024.04 | 16.6 | 0.836 | 1.086 | 16.4 | 18.3 | 13.6 | 87 | |
| C3D-AVG2024.04 | 15.7 | 0.843 | 1.032 | 16.4 | 18.6 | 13.6 | 89 | |
| MSCADC2024.04 | 7.4 | 0.797 | 1.601 | 16.7 | 18.4 | 13.3 | 84 |