Video Question Answering on Egocentric Target Ego-in-Exo Perception EgoSchema
81.28Action Understanding ScoreVISCOP
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| VISCOP2025.10 | 81.28 | 82.8 | 78.75 | 64.86 | 62.11 | 73.96 | 3.53 | |
| VE + Connector FinetuningVE Adaptation Strategy=Trainable, LLM Adaptation Strategy=Frozen, Vision Language Connector Finetuning=Trainable2025.10 | 76.13 | 82.93 | 73.32 | 64.86 | 61.14 | 71.68 | 1.24 | |
| Base VLM2025.10 | 75.37 | 74.88 | 75.56 | 65.38 | 60.98 | 70.43 | — | |
| LLM-LoRA + Connector FinetuningVE Adaptation Strategy=Frozen, LLM Adaptation Strategy=LoRA, Vision Language Connector Finetuning=Trainable2025.10 | 73.49 | 74.27 | 74.5 | 64.99 | 61.52 | 69.75 | -0.68 | |
| Full FinetuningVE Adaptation Strategy=Trainable, LLM Adaptation Strategy=Trainable, Vision Language Connector Finetuning=Trainable2025.10 | 73.28 | 82.68 | 72.96 | 65.77 | 60.31 | 71 | 0.57 | |
| Connector FinetuningVE Adaptation Strategy=Frozen, LLM Adaptation Strategy=Frozen, Vision Language Connector Finetuning=Trainable2025.10 | 73 | 76.71 | 72.85 | 65.51 | 60.43 | 69.7 | -0.74 |