GUI Grounding on ScreenSpot 1.0 (full)
0.78Mobile Text AccSeeClick
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| SeeClickModel Size=9.6B, GUI Specific=true2024.01 | 0.78 | 0.52 | 0.722 | 0.3 | 0.557 | 0.325 | 0.534 | |
| CogAgentModel Size=18B, GUI Specific=true2024.01 | 0.67 | 0.24 | 0.742 | 0.2 | 0.704 | 0.286 | 0.474 | |
| FuyuModel Size=8B, GUI Specific=true2024.01 | 0.41 | 0.013 | 0.33 | 0.036 | 0.339 | 0.044 | 0.195 | |
| GPT-4VModel Size=N/A, GUI Specific=false2024.01 | 0.226 | 0.245 | 0.202 | 0.118 | 0.092 | 0.088 | 0.162 | |
| Qwen-VLModel Size=9.6B, GUI Specific=false2024.01 | 0.095 | 0.048 | 0.057 | 0.05 | 0.035 | 0.024 | 0.052 | |
| MiniGPT-v2Model Size=7B, GUI Specific=false2024.01 | 0.084 | 0.066 | 0.062 | 0.029 | 0.065 | 0.034 | 0.057 |