Fig. 9
From: Heterogeneous dual-decoder network for road extraction in remote sensing images
Model visualization results under different model configurations. White: true positive, red: false negative, green: false positive. Images in rows 1,2 are from the DeepGlobe dataset, images in rows 3,4 are from the Ottawa dataset, and images in rows 5,6 are from the CHN6-CUG dataset. In the figure (a): baseline+Add (b): baseline+DSGD+Add (c): baseline+MCBE+Add (d): baseline+DSGD+MCBE+Add (e): baseline+DSGD+MCBE+Concat (f): HDDNet.
