Replies: 1 comment
-
that's a good point, to add some randomness to the foreground cropping locations, I think you can try |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hi, I'm wondering what cropping method I should use in my case that foreground region is very small. I observe that if I use RandCropByPosNegLabel, many samples will have no foreground. I'm not sure about RandCropByLabelClasses. I think the cropped foreground should not always be centered because the sliding window inference crop would not always get foreground at the center?
If there are any tutorial on how to implement own version of Cropping transform? I would like to implement an improved version based on my special case.
Beta Was this translation helpful? Give feedback.
All reactions