Image-level failure detection on Cityscapes (val)
11.58MAESynthCP-separate
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| SynthCP-separateSegmentation Model=FCN-82020.03 | 11.58 | 11.5 | 64.63 | 65.63 | |
| SynthCP-jointSegmentation Model=FCN-82020.03 | 12.69 | 11.29 | 62.52 | 61.23 | |
| Direct PredictionSegmentation Model=FCN-82020.03 | 13.25 | 11.96 | 58.34 | 59.74 | |
| SynthCP-separateSegmentation Model=Deeplab-v22020.03 | 13.6 | 12.32 | 62.51 | 63.41 | |
| SynthCP-jointSegmentation Model=Deeplab-v22020.03 | 13.68 | 11.6 | 64.05 | 65.42 | |
| Direct PredictionSegmentation Model=Deeplab-v22020.03 | 14.45 | 12.2 | 60.94 | 62.01 | |
| VAE alarmSegmentation Model=FCN-82020.03 | 16.28 | 11.88 | 21.82 | 18.26 | |
| VAE alarmSegmentation Model=Deeplab-v22020.03 | 16.78 | 12.21 | 17.92 | 19.63 | |
| MCDropoutSegmentation Model=FCN-82020.03 | 17.28 | 13.33 | 3.62 | 5.97 | |
| MCDropoutSegmentation Model=Deeplab-v22020.03 | 19.31 | 12.86 | 4.55 | 1.37 |