Multi-choice Video Question Answering on EgoSchema (test)
72.2AccuracyGPT4-O
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT4-OVideo-Gen=false2024.12 | 72.2 | |
| LongVU-7B2025.07 | 67.6 | |
| VideoLLaMA2LLM size=72B, Video-Gen=false2024.12 | 63.9 | |
| Gemini 1.5 ProVideo-Gen=false2024.12 | 63.2 | |
| LLaVA-OneVisionLLM size=72B, Video-Gen=false2024.12 | 62 | |
| LLaVA-NeXT-VideoLLM size=32B, Video-Gen=false2024.12 | 60.9 | |
| LLaVA-OV-7B2025.07 | 60.1 | |
| VILA-40B2025.07 | 58 | |
| LLaVA-Video-7B2025.07 | 57.3 | |
| Gemini 1.0 ProVideo-Gen=false2024.12 | 55.7 | |
| GPT4-VVideo-Gen=false2024.12 | 55.6 | |
| VideoLLaMA2LLM size=7B, Video-Gen=false2024.12 | 51.7 | |
| MoReVQAVLM=PALI-3 (5B), LLM=PaLM-2, Video context=16 uniformly sampled frames2024.04 | 51.7 | |
| InternVL2.5-8B-ITG-32Frames Sampled=322025.07 | 51.6 | |
| InternVL2.5-8B2025.07 | 51.5 | |
| VideoLLaMA2-7B2025.07 | 50.5 | |
| JCEFVLM=PALI-3 (5B), LLM=PaLM-2, Video context=1 frame per second2024.04 | 49.9 | |
| ViperGPT+VLM=PALI-3 (5B), LLM=PaLM-22024.04 | 49.3 | |
| Divot-LLMLLM size=7B, Video-Gen=true2024.12 | 46.5 | |
| LLaVA-NeXT-VideoLLM size=7B, Video-Gen=false2024.12 | 43.9 | |
| VideoChat2LLM size=7B, Video-Gen=false2024.12 | 42.2 | |
| LLM-onlyLLM=PaLM-22024.04 | 41 | |
| LLaMA-VIDLLM size=7B, Video-Gen=false2024.12 | 38.5 | |
| Video-LLaVALLM size=7B, Video-Gen=false2024.12 | 38.4 | |
| Video-LaVITLLM size=7B, Video-Gen=true2024.12 | 37.3 | |
| Random (for MC)2024.04 | 20 |