Localized Relational Captioning on GAR-Bench Cap
62.2Overall ScoreGAR-8B
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| GAR-8BModel Type=Region MLLMs2025.10 | 62.2 | 66 | 64.5 | |
| Gemini-2.5-ProModel Type=Private General MLLMs2025.10 | 59.3 | 51.6 | 66.4 | |
| GAR-1BModel Type=Region MLLMs2025.10 | 57.5 | 56.7 | 63.6 | |
| o3Model Type=Private General MLLMs2025.10 | 56.9 | 37.1 | 74.8 | |
| GPT-4oModel Type=Private General MLLMs2025.10 | 51.5 | 39.2 | 62.6 | |
| Sa2VA-8BModel Type=Region MLLMs2025.10 | 45.6 | 46.4 | 44.9 | |
| InternVL3-38BModel Type=Public General MLLMs2025.10 | 45.1 | 29.9 | 58.9 | |
| Qwen2.5-VL-32BModel Type=Public General MLLMs2025.10 | 36.8 | 17.5 | 54.3 | |
| InternVL3-8BModel Type=Public General MLLMs2025.10 | 33.8 | 11.3 | 54.2 | |
| Qwen2.5-VL-7BModel Type=Public General MLLMs2025.10 | 32.4 | 12.4 | 50.5 | |
| VP-SPHINX-13BModel Type=Region MLLMs2025.10 | 32.3 | 27.8 | 39.3 | |
| InternVL3-2BModel Type=Public General MLLMs2025.10 | 29.4 | 14.4 | 43 | |
| Qwen2.5-VL-3BModel Type=Public General MLLMs2025.10 | 22.5 | 9.3 | 34.6 | |
| PAM-3BModel Type=Region MLLMs2025.10 | 21.1 | 3.1 | 39.3 | |
| DAM-3BModel Type=Region MLLMs2025.10 | 13.1 | 17.5 | 10.3 |