DFormer: Rethinking RGBD Representation Learning for Semantic Segmentation

Benchmark Model Rank Results
rgb-d-salient-object-detection-on-desDFormer-L#1S-Measure: 94.8Average MAE: 0.013max E-Measure: 98.0
rgb-d-salient-object-detection-on-nju2kDFormer-L#1S-Measure: 93.7Average MAE: 0.023max E-Measure: 96.4
rgb-d-salient-object-detection-on-nlprDFormer-L#1S-Measure: 94.2Average MAE: 0.016max E-Measure: 97.1
rgb-d-salient-object-detection-on-sipDFormer-L#1S-Measure: 91.5Average MAE: 0.032max E-Measure: 95.0
rgb-d-salient-object-detection-on-stereDFormer-L#1S-Measure: 92.3Average MAE: 0.030max E-Measure: 95.2
semantic-segmentation-on-nyu-depth-v2DFormer-L#12Mean IoU: 57.2%
semantic-segmentation-on-nyu-depth-v2DFormer-B#21Mean IoU: 55.6%
semantic-segmentation-on-nyu-depth-v2DFormer-S#29Mean IoU: 53.6%
semantic-segmentation-on-nyu-depth-v2DFormer-T#38Mean IoU: 51.8%
semantic-segmentation-on-sun-rgbdDFormer-L#9Mean IoU: 52.5%
semantic-segmentation-on-sun-rgbdDFormer-B#14Mean IoU: 51.2%
semantic-segmentation-on-sun-rgbdTokenFusion (S)#16Mean IoU: 50.0%
semantic-segmentation-on-sun-rgbdFSFNet#21Mean IoU: 48.8%
semantic-segmentation-on-syn-udtiriDFormer#4IoU: 90.88