DepthFromDefocusWithLearnedOptics
DepthFromDefocusWithLearnedOptics copied to clipboard
stretch_depth function
I notice that stretch_depth function in ./datasets/sceneflow.py is not used, so the training depth maps are rescaled to [0, 1] instead of [1, 5] as mentioned in the paper. I wonder if you will first rescale the depth maps in the datasets to [1, 5] through stretch_depth function when you training the model?