Multi-robot cooperative Visual Question Answering on CoopSR iGibson
0.8611T1 ScoreTSPO
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| TSPOEvaluation Protocol=Keyframe, Base Model=Qwen2.5-VL-7B-Instruct2026.05 | 0.8611 | 0.6756 | 0.4529 | 0.5647 | 0.6928 | 0.6144 | 0.5794 | 0.5811 | |
| Qwen2.5-VL-7BEvaluation Protocol=SFT2026.05 | 0.8 | 0.6577 | 0.541 | 0.6474 | 0.634 | 0.6392 | 0.6368 | 0.6367 | |
| InternVL3-8BEvaluation Protocol=SFT2026.05 | 0.7944 | 0.6935 | 0.476 | 0.6897 | 0.5709 | 0.6261 | 0.5856 | 0.5938 | |
| TSPOEvaluation Protocol=Keyframe, Base Model=InternVL3-8B2026.05 | 0.7944 | 0.7202 | 0.4954 | 0.6313 | 0.6221 | 0.6467 | 0.6255 | 0.6312 | |
| SP-CoRBase Model=Qwen3-VL-8B-Instruct2026.05 | 0.7944 | 0.7277 | 0.5942 | 0.7318 | 0.7237 | 0.7029 | 0.695 | 0.707 | |
| EgoRAGEvaluation Protocol=RAG, Base Model=Qwen2.5-VL-7B-Instruct2026.05 | 0.7778 | 0.6815 | 0.5441 | 0.6402 | 0.6194 | 0.6414 | 0.6403 | 0.637 | |
| SP-CoRBase Model=InternVL3-8B2026.05 | 0.7611 | 0.7381 | 0.5517 | 0.7351 | 0.7228 | 0.6832 | 0.6959 | 0.7007 | |
| SP-CoRBase Model=Qwen2.5-VL-7B-Instruct2026.05 | 0.7444 | 0.7188 | 0.5836 | 0.7463 | 0.7274 | 0.6992 | 0.6985 | 0.7082 | |
| Qwen2.5-VL-72BEvaluation Protocol=SFT2026.05 | 0.7389 | 0.7039 | 0.4909 | 0.6024 | 0.5819 | 0.6317 | 0.6116 | 0.6083 | |
| EgoRAGEvaluation Protocol=RAG, Base Model=InternVL3-8B2026.05 | 0.7222 | 0.6905 | 0.4498 | 0.5841 | 0.5691 | 0.6082 | 0.583 | 0.5865 | |
| VideoRAGEvaluation Protocol=RAG, Base Model=Qwen2.5-VL-7B-Instruct2026.05 | 0.6778 | 0.7068 | 0.5182 | 0.5786 | 0.5801 | 0.6026 | 0.6108 | 0.598 | |
| Qwen3-VL-8B-InstructEvaluation Protocol=SFT2026.05 | 0.5944 | 0.5744 | 0.4833 | 0.5469 | 0.6595 | 0.579 | 0.5549 | 0.5424 | |
| Qwen2.5-VL-7B-InstructEvaluation Protocol=Zero-shot2026.05 | 0.3944 | 0.5089 | 0.2812 | 0.2921 | 0.3138 | 0.3561 | 0.3484 | 0.3395 | |
| Qwen3-VL-8B-InstructEvaluation Protocol=Zero-shot2026.05 | 0.3556 | 0.4772 | 0.3766 | 0.3554 | 0.376 | 0.4021 | 0.3533 | 0.3846 | |
| InternVL3-8BEvaluation Protocol=Zero-shot2026.05 | 0.2815 | 0.5225 | 0.348 | 0.3634 | 0.4207 | 0.3861 | 0.3823 | 0.3884 | |
| KTVEvaluation Protocol=Keyframe, Base Model=Qwen2.5-VL-7B-Instruct2026.05 | 0.2167 | 0.5312 | 0.4347 | 0.4203 | 0.3898 | 0.4555 | 0.4579 | 0.4346 | |
| LLaVA-NeXT-Video-7BEvaluation Protocol=Zero-shot2026.05 | 0.2111 | 0.3095 | 0.3237 | 0.3132 | 0.279 | 0.3308 | 0.3171 | 0.3127 | |
| VideoLLaMA3-7BEvaluation Protocol=Zero-shot2026.05 | 0.1944 | 0.4628 | 0.3921 | 0.4009 | 0.3751 | 0.403 | 0.4222 | 0.4005 | |
| Qwen2.5-VL-72B-InstructEvaluation Protocol=Zero-shot2026.05 | 0.1884 | 0.4357 | 0.1982 | 0.3361 | 0.2745 | 0.3018 | 0.3318 | 0.3093 | |
| LLaVA-NeXT-Video-7BEvaluation Protocol=SFT2026.05 | 0.1833 | 0.5327 | 0.4757 | 0.4925 | 0.5619 | 0.5091 | 0.4842 | 0.4805 | |
| VideoLLaMA3-2BEvaluation Protocol=SFT2026.05 | 0.1556 | 0.5312 | 0.4468 | 0.5192 | 0.4511 | 0.5005 | 0.51 | 0.4875 | |
| VideoLLaMA3-7BEvaluation Protocol=SFT2026.05 | 0.1389 | 0.5595 | 0.4772 | 0.5125 | 0.4355 | 0.4361 | 0.5126 | 0.4947 | |
| KTVEvaluation Protocol=Keyframe, Base Model=InternVL3-8B2026.05 | 0.1389 | 0.5863 | 0.4088 | 0.4325 | 0.4227 | 0.4658 | 0.4414 | 0.4431 | |
| VideoLLaMA3-2BEvaluation Protocol=Zero-shot2026.05 | 0.1278 | 0.5551 | 0.3161 | 0.4264 | 0.409 | 0.4067 | 0.4266 | 0.4144 | |
| VideoRAGEvaluation Protocol=RAG, Base Model=InternVL3-8B2026.05 | 0.1222 | 0.6414 | 0.459 | 0.5053 | 0.4794 | 0.5239 | 0.5056 | 0.5029 |