Video Question Answering on EgoSchema 3 min (test)
66.2AccuracyVideoTree
Evaluation Results
| Method | Links | |
|---|---|---|
| VideoTreeVenue=CVPR’25, LLM Size=GPT-4, Frames=63.22025.08 | 66.2 | |
| Qwen2-VL w/ OursVenue=-, LLM Size=7B, Frames=Avg 92025.08 | 65.6 | |
| LLaVA-OV w/ OursVenue=-, LLM Size=7B, Frames=Avg 92025.08 | 65.2 | |
| Qwen2.5-VL w/ OursVenue=-, LLM Size=7B, Frames=Avg 92025.08 | 64.4 | |
| BOLTVenue=CVPR’25, LLM Size=7B, Frames=162025.08 | 61.8 | |
| LLaVA-OV‡Venue=ArXiv’24, LLM Size=7B, Frames=162025.08 | 61.4 | |
| Qwen2-VL‡Venue=ArXiv’24, LLM Size=7B, Frames=162025.08 | 61.2 | |
| VideoAgentVenue=ECCV’24, LLM Size=GPT-4, Frames=8.42025.08 | 60.2 | |
| MVUVenue=ICLR’25, LLM Size=7B, Frames=162025.08 | 60.2 | |
| Qwen2.5-VL‡Venue=ArXiv’25, LLM Size=7B, Frames=162025.08 | 60.2 | |
| Qwen2-VL‡Venue=ArXiv’24, LLM Size=7B, Frames=82025.08 | 59.6 | |
| LLoViVenue=EMNLP’24, LLM Size=GPT-3.5, Frames=902025.08 | 57.6 | |
| Qwen2.5-VL‡Venue=ArXiv’25, LLM Size=7B, Frames=82025.08 | 56.4 | |
| VideoChat2Venue=CVPR’24, LLM Size=7B, Frames=162025.08 | 54.4 | |
| VideoLLaMaVenue=ICCV’25, LLM Size=7B, Frames=82025.08 | 53.8 | |
| VideoLLaMA2Venue=ArXiv’24, LLM Size=7B, Frames=82025.08 | 53.1 | |
| PLLaVAVenue=ArXiv’24, LLM Size=7B, Frames=162025.08 | 45.6 | |
| VideoLLaVAVenue=EMNLP’24, LLM Size=7B, Frames=82025.08 | 40.2 |