PolyMaX: General Dense Prediction with Mask Transformer

Benchmark Model Rank Results
monocular-depth-estimation-on-nyu-depth-v2PolyMaX(ConvNeXt-L)#17absolute relative error: 0.067RMSE: 0.25log 10: 0.029
semantic-segmentation-on-nyu-depth-v2PolyMaX(ConvNeXt-L)#8Mean IoU: 58.08%
surface-normals-estimation-on-nyu-depth-v2-1PolyMaX(ConvNeXt-L)#2% < 11.25: 65.66% < 22.5: 82.28% < 30: 87.83