Multimodal Referring and Grounding on Ferret-Bench v1.0 (test)
79.9Referring DescriptionFerret-v2-7B
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Ferret-v2-7BParameter Size=7B2024.04 | 79.9 | 81.7 | 65.2 | 75.6 | |
| Ferret-v2-13BParameter Size=13B2024.04 | 79.6 | 79.4 | 65.7 | 74.9 | |
| Osprey-7BParameter Size=7B2024.04 | 72.2 | 67.8 | — | — | |
| Ferret-13BParameter Size=13B2024.04 | 70.6 | 68.7 | 59.7 | 66.3 | |
| Ferret-7BParameter Size=7B2024.04 | 68.7 | 67.3 | 57.5 | 64.5 | |
| CogVLM-17BParameter Size=17B2024.04 | 67.1 | 67.6 | 51.7 | 62.1 | |
| SPHINX-2kResolution=2k2024.04 | 55.6 | 70.2 | 66.4 | 64 | |
| Kosmos-22024.04 | 51.8 | 33.7 | 48.4 | 44.6 | |
| Shikra-7BParameter Size=7B2024.04 | 46 | 41.6 | 50.1 | 45.9 | |
| LLaVA2024.04 | 41.4 | 31.7 | 28.8 | 34 |