Region-Level Captioning on refCOCOg (test)
168.2CIDErTraceVision
Evaluation results
| Method | Links | ||
|---|---|---|---|
| TraceVisionParameters=7B2026.02 | 168.2 | 28.8 | |
| VP-SPHINXParameters=13B2026.02 | 162.5 | 23.9 | |
| VP-LLaVAParameters=8B2026.02 | 153.6 | 22.4 | |
| PAMParameters=3B2026.02 | 143.1 | 26.9 | |
| VP-SPHINXParameters=7B2026.02 | 138.7 | 21 | |
| RegionGPTParameters=7B2026.02 | 109.9 | 16.9 | |
| Omni-RGPTParameters=7B2026.02 | 109.7 | 17 | |
| OspreyParameters=7B2026.02 | 108.3 | 16.6 | |
| GLaMMEvaluation protocol=Fine-Tuned (FT)2023.11 | 106 | 16.2 | |
| GLaMMEvaluation protocol=Zero-Shot (ZS)2023.11 | 104 | 15.7 | |
| GLaMMParameters=7B2026.02 | 104 | 15.7 | |
| PixelLLM2026.02 | 82.3 | 14.3 | |
| GRiT2023.11 | 71.6 | 15.2 | |
| MMFuserBase Model=LLaVA-1.5-13B2024.10 | 63.4 | — | |
| Kosmos-22023.11 | 62.3 | 14.1 | |
| MMFuserBase Model=LLaVA-1.5-7B2024.10 | 61.6 | — | |
| LLaVA-1.5-13BParameters=13B2024.10 | 59.9 | — | |
| LLaVA-1.5-7BParameters=7B2024.10 | 58.9 | — |