Spatial Visual Question Answering on Human-annotated quantitative spatial relationship dataset (test)
99Output Accuracy (%)SpatialVLM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SpatialVLM2024.01 | 99 | 37.2 | |
| PaLM 2-E2024.01 | 88.8 | 33.9 | |
| PaLM-E2024.01 | 83.2 | 23.7 | |
| PaLI2024.01 | 52 | 5.3 | |
| InstructBLIP2024.01 | 26 | 7.9 | |
| LLaVA-1.52024.01 | 20.9 | 13 | |
| GPT-4V2024.01 | 1 | 0 |