Dense Captioning on VG v1.2 (Metrics C and P)
31.9Metric CPatch-ioner
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Patch-ionerConfiguration=T2D + Mem., Number of Parameters=0.21B, Adaptation strategy=Patch-based2025.10 | 31.9 | 78.8 | |
| Llava1.5 One-VisionAdaptation strategy=Visual Prompting, Number of Parameters=4B2025.10 | 25.3 | 76.2 | |
| Qwen2.5 VLAdaptation strategy=Visual Prompting, Number of Parameters=3B2025.10 | 19.1 | 74.2 | |
| Llava1.5 One-VisionAdaptation strategy=Crop, Number of Parameters=4B2025.10 | 15.1 | 72.7 | |
| Qwen2.5 VLAdaptation strategy=Crop, Number of Parameters=3B2025.10 | 11.9 | 69.2 | |
| Qwen3 VLAdaptation strategy=Visual Prompting, Number of Parameters=4B2025.10 | 10.8 | 74.2 | |
| Qwen3 VLAdaptation strategy=Crop, Number of Parameters=4B2025.10 | 3.5 | 68.8 |