Embodied Question Answering on AVS-ProcTHOR
93.02Existence AccuracyTarget view
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Target viewAction Model Category=No Action, Answer Verifier=Gemini-2.5-Flash2026.07 | 93.02 | 69.14 | 92.58 | 84.91 | |
| SFT+RLAction Model Category=VG-AVS Framework, Answer Verifier=Gemini-2.5-Flash2026.07 | 91.47 | 69.52 | 90.17 | 83.72 | |
| SFTAction Model Category=VG-AVS Framework, Answer Verifier=Gemini-2.5-Flash2026.07 | 91.28 | 57.06 | 83.84 | 77.39 | |
| RLAction Model Category=VG-AVS Framework, Answer Verifier=Gemini-2.5-Flash2026.07 | 86.82 | 65.24 | 83.41 | 78.49 | |
| Gemini-2.5-ProAction Model Category=Proprietary Models, Answer Verifier=Gemini-2.5-Flash2026.07 | 82.95 | 52.79 | 81 | 72.25 | |
| GPT-5Action Model Category=Proprietary Models, Answer Verifier=Gemini-2.5-Flash2026.07 | 81.01 | 55.58 | 79.69 | 72.09 | |
| LIMEAction Model Category=Ours, Answer Verifier=Gemini-2.5-Flash2026.07 | 65.5 | 36.62 | 72.71 | 57.41 | |
| Qwen2.5-VL-7BAction Model Category=Backbone Model, Answer Verifier=Gemini-2.5-Flash2026.07 | 64.34 | 29.74 | 56.55 | 50.21 | |
| Fine-EQAAction Model Category=EQA Framework, Answer Verifier=Gemini-2.5-Flash2026.07 | 63.57 | 31.97 | 64.41 | 53.32 | |
| ViLaSRAction Model Category=Spatial VLMs, Answer Verifier=Gemini-2.5-Flash2026.07 | 57.95 | 25.46 | 52.84 | 45.42 | |
| SpatialReasonerAction Model Category=Spatial VLMs, Answer Verifier=Gemini-2.5-Flash2026.07 | 54.65 | 22.68 | 52.62 | 43.32 | |
| Query viewAction Model Category=No Action, Answer Verifier=Gemini-2.5-Flash2026.07 | 49.22 | 16.36 | 61.57 | 42.38 |