Referring Object Detection on RefCOCO low-data (val)
82.51Top-1 AccuracyHeROD-G
Evaluation results
| Method | Links | |
|---|---|---|
| HeROD-GTraining data percentage=5%, Backbone=Swin-T, Detector=DINO2026.03 | 82.51 | |
| HeROD-GTraining data percentage=2%, Backbone=Swin-T, Detector=DINO2026.03 | 79.43 | |
| Grounding DINOTraining data percentage=5%, Backbone=Swin-T, Detector=DINO2026.03 | 79.32 | |
| HeROD-GTraining data percentage=1%, Backbone=Swin-T, Detector=DINO2026.03 | 77.91 | |
| HeROD-GTraining data percentage=0.5%, Backbone=Swin-T, Detector=DINO2026.03 | 74.32 | |
| HeROD-GTraining data percentage=0.2%, Backbone=Swin-T, Detector=DINO2026.03 | 73.37 | |
| HeROD-GTraining data percentage=0.1%, Backbone=Swin-T, Detector=DINO2026.03 | 70.82 | |
| Grounding DINOTraining data percentage=2%, Backbone=Swin-T, Detector=DINO2026.03 | 66.6 | |
| HeROD-UTraining data percentage=5%, Backbone=ResNet-50, Detector=Deformable DETR2026.03 | 65.33 | |
| Grounding DINOTraining data percentage=1%, Backbone=Swin-T, Detector=DINO2026.03 | 63.66 | |
| Grounding DINOTraining data percentage=0.5%, Backbone=Swin-T, Detector=DINO2026.03 | 60.97 | |
| HeROD-UTraining data percentage=2%, Backbone=ResNet-50, Detector=Deformable DETR2026.03 | 59.45 | |
| Grounding DINOTraining data percentage=0.2%, Backbone=Swin-T, Detector=DINO2026.03 | 59.17 | |
| Grounding DINOTraining data percentage=0.1%, Backbone=Swin-T, Detector=DINO2026.03 | 57.93 | |
| UNINEXTTraining data percentage=5%, Backbone=ResNet-50, Detector=Deformable DETR2026.03 | 54.3 | |
| HeROD-UTraining data percentage=1%, Backbone=ResNet-50, Detector=Deformable DETR2026.03 | 47.89 | |
| UNINEXTTraining data percentage=2%, Backbone=ResNet-50, Detector=Deformable DETR2026.03 | 46.24 | |
| HeROD-UTraining data percentage=0.5%, Backbone=ResNet-50, Detector=Deformable DETR2026.03 | 35.75 | |
| HeROD-UTraining data percentage=0.2%, Backbone=ResNet-50, Detector=Deformable DETR2026.03 | 31.59 | |
| UNINEXTTraining data percentage=1%, Backbone=ResNet-50, Detector=Deformable DETR2026.03 | 31.04 | |
| HeROD-UTraining data percentage=0.1%, Backbone=ResNet-50, Detector=Deformable DETR2026.03 | 25.6 | |
| UNINEXTTraining data percentage=0.5%, Backbone=ResNet-50, Detector=Deformable DETR2026.03 | 24.05 | |
| UNINEXTTraining data percentage=0.2%, Backbone=ResNet-50, Detector=Deformable DETR2026.03 | 21.53 | |
| UNINEXTTraining data percentage=0.1%, Backbone=ResNet-50, Detector=Deformable DETR2026.03 | 17.75 |