CRF in deeplab
Traditionally, conditional random fields (CRFs) have been employed to smooth noisy segmentation maps.
Typically these models couple neighboring nodes, favoring same-label assignments to spatially proximal pixels. Qualitatively, the primary function of these short-range CRFs is to clean up the spurious predictions of weak classi- fiers built on top of local hand-engineered features.
The score maps are typically quite smooth and produce homogeneous classification results. In this regime, using short-range CRFs can be detrimental, as our goal should be to recover detailed local structure rather than further smooth it.