End-to-end Training of CNN-CRF via Differentiable Dual-Decomposition

Re-assign community