Various doc fixes for transforms #3704
Conversation
torchvision/transforms/functional.py
Outdated
|
|
||
| - constant: pads with a constant value, this value is specified with fill | ||
|
|
||
| - edge: pads with the last value on the edge of the image, | ||
| if input a 5D torch Tensor, the last 3 dimensions will be padded instead of the last 2 | ||
| - edge: pads with the last value at the edge of the image |
There was a problem hiding this comment.
With the propsoed changes the docs look like this:
Perhaps it would look slightly better like this:
Which requires the next line starting on the same column of the e of edge, i.e.:
- edge: pads with the last value at the edge of the image.
If input a 5D torch Tensor, the last 3 dimensions will be padded instead of the last 2
- reflect: pads with reflection of image without repeating the last value on the edge.
For example, padding [1, 2, 3, 4] with 2 elements on both sides in reflect mode
will result in [3, 2, 1, 2, 3, 4, 3, 2]
- symmetric: pads with reflection of image repeating the last value on the edge.
For example, padding [1, 2, 3, 4] with 2 elements on both sides in symmetric mode
will result in [2, 1, 1, 2, 3, 4, 4, 3]
(You'll need to add full dots at the end of the sentences otherwise it will not read well).
WDYT?
There was a problem hiding this comment.
@NicolasHug Yes I think your suggestion does look better (tighter but smoother), I'll commit again in a bit.
There was a problem hiding this comment.
Changed as requested 87c4885
NicolasHug
changed the title
|
On second thought, the |
|
It's the oldest trick in the book: introduce some subtle bugs so you can claim the fixes later 😆 |
Reviewed By: NicolasHug Differential Revision: D28169133 fbshipit-source-id: 3a6c789b88263f2ed01c6e5ffba6ad8a972d52de
padding_modebullet listsJust found these when building master docs yesterday, what do you think? @datumbox