Multiple Choice Video Question Answering on VMB
53Spatial Accuracy (S)VideoMindPalace
Evaluation results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| VideoMindPalaceLLM=GPT-42025.01 | 53 | 51.5 | 49.1 | 51.2 | 57.2 | 56.6 | 55.1 | 56.3 | 37 | 33.7 | 32.8 | 34.5 | |
| VideoAgentLLM=GPT-42025.01 | 50.5 | 45.9 | 44.7 | 47 | 58.2 | 55.3 | 51.2 | 54.9 | 25 | 24.7 | 23.8 | 24.5 | |
| LLoViLLM=GPT-42025.01 | 44.5 | 41.8 | 40.6 | 42.3 | 55.2 | 53.1 | 50.1 | 52.8 | 23.1 | 22.3 | 21.8 | 22.4 | |
| VideoTreeLLM=GPT-42025.01 | 43.2 | 40.8 | 37.5 | 40.5 | 47 | 45.5 | 43.4 | 45.3 | 23.9 | 22.6 | 22.2 | 22.9 | |
| IG-VLMLLM=GPT-42025.01 | 32.8 | 31.2 | 30.5 | 31.5 | 36.2 | 33.8 | 30.5 | 33.5 | 22.4 | 21.9 | 21.1 | 21.8 | |
| Random2025.01 | 20 | 20 | 20 | 20 | 20 | 20 | 20 | 20 | 20 | 20 | 20 | 20 |