Fig. 3From: Deep residual coalesced convolutional network for efficient semantic road segmentationResidual coalesced convolutional modulesBack to article page