In this paper, we propose a novel model and its training strategy for the CRF.

This section introduces papers related to our work.
* structured output CNN (citation not found: Yann)

* CNN CRF (citation not found: Zisserman)

* Structured loss CNN (citation not found: Honglak)

* Other papers (citation not found: chen14iclr), (Tompson 2014)

* CRF-RNN (citation not found: zheng15iccv)

* DeepLab (citation not found: nothing)

* Deep parsing network (Liu 2015)

* Bilinear (Lin 2015)

We denote the input image by \(I\), output by \(Y\), set of parameters by \(\Theta\).

Conditional Random Fields(CRF).

\begin{equation} S(I,Y,\Theta)=\sum_{i\in V}{\Phi_{i}(I,Y_{i},\Theta)+\sum_{ij\in E}{\Psi_{ij}(I,Y_{i},Y_{j},\Theta)}}\\ \end{equation}
\begin{equation}
\Psi_{ij}(I,Y_{i},Y_{j},\Theta)=\sum_{m=1}^{M}{w_{m}k_{m}(f(Y_{i}),f(Y_{j}))}\\
\end{equation}

\begin{equation}
Y^{*}=\operatorname*{arg\,max}_{Y}{S(I,Y,\Theta)}\\
\end{equation}