• CRFs

learning

For example, it
penalizes a pair of nearby pixels labeled “sky” and “bird” to the same extent as pixels labeled “sky”
and “cat”. We can instead learn a general symmetric compatibility function µ(xi
, xj ) that takes
interactions between labels into account, as described in Section 4

Inference

• smoothness kernel removes small isolated regions

• The appearance kernel is inspired by the observation that nearby pixels with similar color are likely to be in the same class