Video Understanding on Ego-in-Exo, NeXTQA, Video MME, ADL-X
71.89Ego-in-Exo (Exo RGB) ScoreVISCOP
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| VISCOPVision Language Connector=trainable2025.10 | 71.89 | 83.91 | 64.3 | 76.59 | 76.47 | 74.63 | 1.84 | |
| Finetuning (Frozen VE, Frozen LLM)Vision Encoder=frozen, LLM=frozen, Vision Language Connector=trainable2025.10 | 71.36 | 83.15 | 62.41 | 70.9 | 69.05 | 71.37 | -1.42 | |
| Base VLMVision Encoder=frozen, LLM=frozen2025.10 | 66.27 | 84.32 | 65.37 | 77.36 | 70.65 | 72.79 | — | |
| Finetuning (Trainable VE, Frozen LLM)Vision Encoder=trainable, LLM=frozen, Vision Language Connector=trainable2025.10 | 60.97 | 82.89 | 62 | 71.48 | 67.26 | 68.92 | -3.87 | |
| Finetuning (Frozen VE, LoRA LLM)Vision Encoder=frozen, LLM=LoRA updates, Vision Language Connector=trainable2025.10 | 60.97 | 83.73 | 64.19 | 72.19 | 72.49 | 70.71 | -2.08 |