Closed abderhasan closed 5 years ago
I have some ground truth images. In order to make them work with UNet, should I label each pixel (i.e. for semantic segmentation), or some binary ground truth for 2 classes (black: background, white: foreground) would work?
Thanks.
Is this a duplicate to #223 ?
I have some ground truth images. In order to make them work with UNet, should I label each pixel (i.e. for semantic segmentation), or some binary ground truth for 2 classes (black: background, white: foreground) would work?
Thanks.