Temporal Grounding on EgoExo-Con 1.0 (test)
72.4G-ExoHuman
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| HumanInput frames=-, Trained on ego data=-2025.10 | 72.4 | 73 | 67.3 | |
| Gemini-2.5 FlashInput frames=1 fps, Trained on ego data=-2025.10 | 42 | 45.9 | 20.8 | |
| GPT-5Input frames=32, Trained on ego data=-2025.10 | 34.5 | 32.8 | 20.1 | |
| TimeSuiteInput frames=128, Trained on ego data=✓2025.10 | 28.2 | 27.3 | 18.7 | |
| TimeChat-VTInput frames=96, Trained on ego data=✗2025.10 | 27.8 | 26.2 | 16.3 | |
| Video-LLaMA3Input frames=1 fps, Trained on ego data=✓2025.10 | 27.7 | 28 | 16.2 | |
| TimeChatInput frames=96, Trained on ego data=✗2025.10 | 21.3 | 20.5 | 12.8 | |
| Qwen2.5-VLInput frames=1 fps, Trained on ego data=✗2025.10 | 14.2 | 11.4 | 6.9 | |
| VTimeLLMInput frames=100, Trained on ego data=✗2025.10 | 12.6 | 11.1 | 6.5 | |
| RandomInput frames=-, Trained on ego data=-2025.10 | 12.5 | 12.5 | 12.5 | |
| Video-LLaMA2Input frames=8, Trained on ego data=✓2025.10 | 12 | 11.5 | 7.5 | |
| VideoChat2Input frames=16, Trained on ego data=✓2025.10 | 5.6 | 5.3 | 4 |