Patch-based image classification through conditional random field model

We present an image classification system based on a Conditional Random Field (CRF) model trained on simple features obtained from a small number of semantically representative image patches. The CRFs are very powerful to handle complex parts dependencies due to their approach based on the effective... View More