3D Visual Grounding on Nr3D 2 (val)
72.1Accuracy@0.25 (Easy)3D-VisTA
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| 3D-VisTASupervision=Fully Supervised2026.06 | 72.1 | 56.7 | 61.5 | 65.1 | 64.2 | |
| Agentic Collaborative CognitionSupervision=Zero-shot (250 queries)2026.06 | 69.9 | 60.5 | 61.5 | 68.2 | 65.6 | |
| MiKASASupervision=Fully Supervised2026.06 | 69.7 | 59.4 | 65.4 | 64 | 64.4 | |
| SPAZERSupervision=Zero-shot (250 queries)2026.06 | 68 | 58.8 | 59.9 | 66.2 | 63.8 | |
| CSVGSupervision=Zero-shot (full)2026.06 | 67.1 | 51.3 | 53 | 62.5 | 59.2 | |
| Agentic Collaborative CognitionSupervision=Zero-shot (full)2026.06 | 65.7 | 57 | 56.7 | 63.7 | 61.2 | |
| BUTD-DETRSupervision=Fully Supervised2026.06 | 60.7 | 48.4 | 46 | 58 | 54.6 | |
| SeqVLMSupervision=Zero-shot (250 queries)2026.06 | 58.1 | 47.4 | 51 | 54.5 | 53.2 | |
| VLM-GrounderSupervision=Zero-shot (250 queries)2026.06 | 55.2 | 39.5 | 45.8 | 49.4 | 48 | |
| SeeGroundSupervision=Zero-shot (full)2026.06 | 54.5 | 38.3 | 42.3 | 48.2 | 46.1 | |
| 3DVG-TSupervision=Fully Supervised2026.06 | 48.5 | 34.8 | 34.8 | 43.7 | 40.8 | |
| ZSVG3DSupervision=Zero-shot (full)2026.06 | 46.5 | 31.7 | 36.8 | 40 | 39 | |
| ReferIt3DNetSupervision=Fully Supervised2026.06 | 43.6 | 27.9 | 32.5 | 37.1 | 35.6 |