Referring Object Detection on RefCOCO (testA)
91.46Top-1 AccuracyGrounding DINO*
Evaluation Results
| Method | Links | |
|---|---|---|
| Grounding DINO*Fine-tuning=w/2026.03 | 91.46 | |
| HeROD-GFine-tuning=w/2026.03 | 91.46 | |
| DQ-DETRFine-tuning=w/2026.03 | 91.04 | |
| MDETRFine-tuning=w/2026.03 | 89.58 | |
| RefTRFine-tuning=w/2026.03 | 88.73 | |
| VILLA-LFine-tuning=w/2026.03 | 87.48 | |
| HeROD-UFine-tuning=w/2026.03 | 84.87 | |
| TransVGFine-tuning=w/2026.03 | 82.72 | |
| VGTRFine-tuning=w/2026.03 | 82.32 | |
| MAttNetFine-tuning=w/2026.03 | 81.14 | |
| UNINEXT*Fine-tuning=w/2026.03 | 80.86 | |
| HeROD-GSupport Data=2k, Stage=Support Set Training2026.03 | 80.5 | |
| HeROD-GSupport Data=2k, Stage=2k Finetune2026.03 | 80.5 | |
| HeROD-GSupport Data=2k, Stage=1k Finetune2026.03 | 80.15 | |
| HeROD-GSupport Data=1k, Stage=Support Set Training2026.03 | 79.51 | |
| HeROD-GSupport Data=1k, Stage=2k Finetune2026.03 | 79.51 | |
| HeROD-GSupport Data=2k, Stage=0.5k Finetune2026.03 | 78.75 | |
| HeROD-GSupport Data=1k, Stage=1k Finetune2026.03 | 78.27 | |
| HeROD-GSupport Data=1k, Stage=0.5k Finetune2026.03 | 76.91 | |
| Grounding DINOSupport Data=2k, Stage=2k Finetune2026.03 | 72.3 | |
| Grounding DINOSupport Data=2k, Stage=Support Set Training2026.03 | 70.66 | |
| Grounding DINOSupport Data=1k, Stage=2k Finetune2026.03 | 69.91 | |
| Grounding DINOSupport Data=2k, Stage=1k Finetune2026.03 | 69.56 | |
| Grounding DINOSupport Data=2k, Stage=0.5k Finetune2026.03 | 68.84 | |
| Grounding DINOSupport Data=1k, Stage=Support Set Training2026.03 | 68.62 | |
| Grounding DINOSupport Data=1k, Stage=1k Finetune2026.03 | 67.93 | |
| Grounding DINOSupport Data=1k, Stage=0.5k Finetune2026.03 | 67.74 | |
| Grounding DINO*Fine-tuning=w/o2026.03 | 57.43 |