Unintuitive behaviour of RandAffine
and RandAffined
#8387
MathijsdeBoer
started this conversation in
General
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
The default behaviour of
RandAffine
andRandAffined
'sscale_range
parameter is to add a value of1.0
to all elements of the argument.See the documentation:
scale_range
– scaling range with format matching rotate_range. it defines the range to randomly select the scale factor to translate for every spatial dims. A value of 1.0 is added to the result. This allows 0 to correspond to no change (i.e., a scaling of 1.0).This, while clearly stated in the documentation, does seem a little unintuitive, and it has in fact caught me out, where I was simply passing
[0.9, 1.1]
to the constructor, causing strange warnings downstream in my random cropping step.Would it be too disruptive to change this to a scale where a value of
1.0
is no change?Beta Was this translation helpful? Give feedback.
All reactions