Urban Perception Dimension Scores on Place Pulse 2.0
7.4CleanlinessUrbanFusion
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| UrbanFusionEvaluation Protocol=Linear Regression2025.10 | 7.4 | 12.5 | 22.7 | 27.1 | 23.7 | 19.1 | 18.8 | |
| GeoCLIPEvaluation Protocol=Linear Regression, Training Dataset=MP-162025.10 | 7.2 | 12.7 | 23.3 | 27.6 | 23.6 | 20.8 | 19.2 | |
| GAIREvaluation Protocol=Linear Regression2025.10 | 6.7 | 11 | 21.1 | 25.7 | 21.2 | 19 | 17.4 | |
| GeoCLIPEvaluation Protocol=MLP, Pre-training Dataset=PP2-M2025.10 | 6.6 | 3.7 | 4.9 | 10.9 | 7 | 7.7 | 4.6 | |
| GeoCLIPEvaluation Protocol=MLP Regression, Training Dataset=PP2-M2025.10 | 5.3 | 3.8 | 9.4 | 7.9 | 6.1 | 6.6 | 4.8 | |
| GeoCLIPEvaluation Protocol=Linear Regression, Training Dataset=PP2-M2025.10 | 5.2 | 9.5 | 19.6 | 22.5 | 19.2 | 17.1 | 15.5 | |
| SatCLIP L40Evaluation Protocol=Linear Regression, Training Dataset=S2-100K (vs PP2-M Models)2025.10 | 3.5 | 6.9 | 11.5 | 14.1 | 10.1 | 10.7 | 9.5 | |
| SatCLIP L10Evaluation Protocol=Linear Regression, Training Dataset=S2-100K (vs MP-16 Models)2025.10 | 3.3 | 6.9 | 12.1 | 13.9 | 10.4 | 10.8 | 9.6 | |
| SatCLIP L10Evaluation Protocol=Linear Regression, Training Dataset=S2-100K (vs PP2-M Models)2025.10 | 3.2 | 6.9 | 11.4 | 14.5 | 10.1 | 11.1 | 9.5 | |
| SatCLIP L40Evaluation Protocol=Linear Regression, Training Dataset=S2-100K (vs MP-16 Models)2025.10 | 3 | 7 | 12.2 | 14 | 10.1 | 10.7 | 9.5 | |
| UrbanFusionEvaluation Protocol=MLP Regression2025.10 | 2.9 | 8.1 | 17.7 | 22.1 | 17.5 | 14.6 | 13.8 | |
| UrbanFusionEvaluation Protocol=Linear Regression, Pre-training Dataset=PP2-M2025.10 | 2.8 | 6.5 | 11.1 | 14.3 | 10.7 | 11.5 | 9.5 | |
| SatCLIP L40Evaluation Protocol=MLP Regression, Training Dataset=S2-100K (vs PP2-M Models)2025.10 | 2.7 | 5.1 | 9.7 | 12.5 | 8.7 | 8.7 | 7 | |
| GeoCLIPEvaluation Protocol=Linear Regression, Pre-training Dataset=MP-162025.10 | 2.5 | 5.3 | 9 | 13 | 8.7 | 10 | 8.1 | |
| GAIREvaluation Protocol=Linear Regression, Pre-training Dataset=PP2-M2025.10 | 2.4 | 5.4 | 8.9 | 12.2 | 8.3 | 9.6 | 7.8 | |
| GeoCLIPEvaluation Protocol=MLP, Pre-training Dataset=MP-162025.10 | 2.1 | 4.1 | 7.2 | 12.3 | 7.2 | 8.3 | 6.2 | |
| GeoCLIPEvaluation Protocol=Linear Regression, Pre-training Dataset=PP2-M2025.10 | 2 | 5.8 | 9.8 | 12.3 | 8.3 | 10 | 8 | |
| GeoCLIPEvaluation Protocol=MLP Regression, Training Dataset=MP-162025.10 | 1.7 | 5.9 | 17.9 | 21.3 | 18.2 | 15.4 | 13.4 | |
| SatCLIP_L40Evaluation Protocol=Linear Regression, Pre-training Dataset=PP2-M2025.10 | 1 | 3.8 | 7 | 11.2 | 5.8 | 7.9 | 6.1 | |
| CSPEvaluation Protocol=Linear Regression, Pre-training Dataset=FMoW2025.10 | 0.9 | 3.1 | 6.4 | 8.4 | 3 | 7 | 4.8 | |
| CSPEvaluation Protocol=Linear Regression, Backbone=FMoW2025.10 | 0.9 | 3.1 | 6.4 | 8.4 | 3 | 7 | 4.8 | |
| CSPEvaluation Protocol=Linear Regression, Pre-training Dataset=iNat2025.10 | 0.8 | 3.1 | 6 | 8.6 | 3.1 | 6.9 | 4.8 | |
| CSPEvaluation Protocol=Linear Regression, Backbone=iNat2025.10 | 0.8 | 3.1 | 6 | 8.6 | 3.1 | 6.9 | 4.8 | |
| SatCLIP_L10Evaluation Protocol=Linear Regression, Pre-training Dataset=PP2-M2025.10 | 0.7 | 3.8 | 6.7 | 9.8 | 3.4 | 7.5 | 5.3 | |
| SatCLIP_L10Evaluation Protocol=Linear Regression, Pre-training Dataset=S2-100K2025.10 | 0.7 | 3.5 | 6.6 | 9.5 | 3.8 | 7.2 | 5.2 | |
| SatCLIP_L40Evaluation Protocol=Linear Regression, Pre-training Dataset=S2-100K2025.10 | 0.7 | 3.8 | 6.5 | 10 | 5.5 | 7.8 | 5.7 | |
| SatCLIP_L10Evaluation Protocol=MLP, Pre-training Dataset=PP2-M2025.10 | 0.6 | 2.7 | 3.7 | 6.7 | 2.4 | 6.1 | 3.7 | |
| CSPEvaluation Protocol=MLP, Pre-training Dataset=FMoW2025.10 | 0.6 | 3.2 | 6.2 | 7 | 2.5 | 5.5 | 4.2 | |
| CSPEvaluation Protocol=MLP Regression, Backbone=FMoW2025.10 | 0.6 | 3.2 | 6.2 | 7 | 2.5 | 5.5 | 4.2 | |
| SatCLIP L10Evaluation Protocol=MLP Regression, Training Dataset=S2-100K (vs PP2-M Models)2025.10 | 0.5 | 6.1 | 11.1 | 14.2 | 10.8 | 10.3 | 8.7 | |
| SatCLIP_L40Evaluation Protocol=MLP, Pre-training Dataset=S2-100K2025.10 | 0.4 | 3.4 | 5.1 | 8 | 2.9 | 5.6 | 4.2 | |
| CSPEvaluation Protocol=MLP, Pre-training Dataset=iNat2025.10 | 0.4 | 2.3 | 5.6 | 7.1 | 2.4 | 5.5 | 3.8 | |
| IdentityEvaluation Protocol=MLP, Source=Google y ~ g(c)2025.10 | 0.4 | 2.8 | 5.5 | 6.8 | 2.6 | 5.2 | 3.9 | |
| CSPEvaluation Protocol=MLP Regression, Backbone=iNat2025.10 | 0.4 | 2.3 | 5.6 | 7.1 | 2.4 | 5.5 | 3.8 | |
| IdentityEvaluation Protocol=MLP Regression, Implementation=Google y ~ g(c)2025.10 | 0.4 | 2.8 | 5.5 | 6.8 | 2.6 | 5.2 | 3.9 | |
| UrbanFusionEvaluation Protocol=MLP, Pre-training Dataset=PP2-M2025.10 | 0 | 2.2 | 6.1 | 10.8 | 6.2 | 7.9 | 5.5 | |
| GAIREvaluation Protocol=MLP, Pre-training Dataset=PP2-M2025.10 | 0 | 1.6 | 5.2 | 10 | 5.4 | 5.9 | 4.7 | |
| SatCLIP_L40Evaluation Protocol=MLP, Pre-training Dataset=PP2-M2025.10 | 0 | 1.5 | 5.7 | 9 | 4.6 | 7.7 | 4.7 | |
| SatCLIP_L10Evaluation Protocol=MLP, Pre-training Dataset=S2-100K2025.10 | 0 | 2.9 | 4.8 | 7.1 | 2.4 | 5.4 | 3.8 | |
| IdentityEvaluation Protocol=Linear Regression, Source=Google y ~ g(c)2025.10 | 0 | 0.7 | 2.3 | 2.2 | 0.4 | 2.1 | 1.3 | |
| IdentityEvaluation Protocol=Linear Regression, Implementation=Google y ~ g(c)2025.10 | 0 | 0.7 | 2.3 | 2.2 | 0.4 | 2.1 | 1.3 | |
| GAIREvaluation Protocol=MLP Regression2025.10 | 0 | 2.9 | 9.9 | 12.5 | 10.3 | 8.7 | 6.4 | |
| SatCLIP L10Evaluation Protocol=MLP Regression, Training Dataset=S2-100K (vs MP-16 Models)2025.10 | 0 | 6.1 | 11.5 | 14.2 | 11 | 10.6 | 8.9 | |
| SatCLIP L40Evaluation Protocol=MLP Regression, Training Dataset=S2-100K (vs MP-16 Models)2025.10 | 0 | 5.9 | 9.6 | 13.5 | 9.8 | 8.7 | 7.9 |