Figure 6.

Qualitative results of the proposed DED-CNN in the cross-dataset setting where all the models are trained on different dataset and tested on different dataset. From left to right: the first row is the input image, the second row represent the ground truths, the third row is the prediction result of the first encode-decoder model in DED-CNN framework and the last row represent the results of the second encoder–decoder model in the DED-CNN framework. All the models are trained and test in cross dataset configuration.