Video Perception on Perception
72.8AccuracyFrameOracle
Evaluation Results
| Method | Links | |
|---|---|---|
| FrameOracleBackbone=VideoLLaMA3-7B, Frames=64→13.92025.10 | 72.8 | |
| VideoLLaMA3-7BFrames=162025.10 | 72.3 | |
| FrameOracleBackbone=VideoLLaMA3-7B, Frames=16→10.42025.10 | 72 | |
| FrameOracleBackbone=Qwen3-VL-8B, Frames=128→27.82025.10 | 69 | |
| FrameOracleBackbone=Qwen3-VL-8B, Frames=32→20.92025.10 | 68.2 | |
| FrameOracleBackbone=Qwen2.5-VL-3B, Frames=128→27.82025.10 | 67.8 | |
| Qwen3-VL-8BFrames=322025.10 | 67.5 | |
| FrameOracleBackbone=Qwen2.5-VL-3B, Frames=32→20.92025.10 | 66.7 | |
| Qwen2.5-VL-3BFrames=322025.10 | 65.9 | |
| FrameOracleBackbone=LLaVA-Video-7B, Frames=64→13.92025.10 | 65.1 | |
| FrameOracleBackbone=LLaVA-Video-7B, Frames=16→10.42025.10 | 64.7 | |
| LLaVA-Video-7BFrames=162025.10 | 64.3 | |
| FrameOracleBackbone=LLaVA-OneVision-7B, Frames=64→13.92025.10 | 56.9 | |
| FrameOracleBackbone=LLaVA-OneVision-7B, Frames=16→10.42025.10 | 56.5 | |
| LLaVA-OneVision-7BFrames=162025.10 | 56.4 | |
| VideoLLaMA2-7BFrames=162025.10 | 54.9 |