Referring Description on RefCOCOg
5.53Recall @ 4LLaVA + ControlMLLM
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| LLaVA + ControlMLLMBackbone=LLaVA, Visual Prompting=Learnable latent variable optimization2024.07 | 5.53 | 14 | 59.75 | 19.08 | |
| LLaVA + ColorBackbone=LLaVA, Visual Prompting=Coloring the region2024.07 | 5.37 | 11.57 | 55.27 | 17.01 | |
| LLaVABackbone=LLaVA2024.07 | 5.02 | 13.15 | 55.61 | 17.61 | |
| InstructBLIP + ControlMLLMBackbone=InstructBLIP, Visual Prompting=Learnable latent variable optimization2024.07 | 1.39 | 8.77 | 10.28 | 8.24 | |
| InstructBLIP + ColorBackbone=InstructBLIP, Visual Prompting=Coloring the region2024.07 | 1.27 | 8.26 | 14.16 | 6.92 | |
| InstructBLIPBackbone=InstructBLIP2024.07 | 1.24 | 8.7 | 9.89 | 7.95 |