Long-form first-person temporal reasoning on EgoSchema (test)
51.2AccuracyLLaVA-Mini
Evaluation Results
| Method | Links | |
|---|---|---|
| LLaVA-Mini#Frames=1fps2025.01 | 51.2 | |
| VideoChat#Frames=162025.01 | 42.2 | |
| LLAMA-VID#Frames=1fps2025.01 | 38.5 | |
| Video-LLaVA#Frames=82025.01 | 38.4 | |
| Video-ChatGPT#Frames=1002025.01 | 36.2 | |
| TimeChat#Frames=962025.01 | 33 | |
| InternVideo#Frames=162025.01 | 32.1 | |
| mPLUG-Owl#Frames=162025.01 | 31.1 | |
| Random2025.01 | 20 |