class
SmallestMaxSize
extends
_MaxSizeResizeSmallestMaxSize(max_size: int, interpolation: int | str | Interpolation = 1, p: float = 1.0)Scale so the shorter side equals max_size (Albumentations SmallestMaxSize).
Computes the scale factor max_size / min(H, W) and applies it to
both axes — the aspect ratio is preserved and the longer side ends
up at least as large as max_size. Image uses interpolation,
masks use nearest, boxes / keypoints scale exactly.
Parameters
max_sizeintTarget length of the shorter side after resize.
Image resampling mode (OpenCV codes accepted).
pfloat= 1.0Probability of applying the transform.
Examples
>>> import lucid, lucid.utils.transforms as T
>>> tf = T.SmallestMaxSize(16, p=1.0)
>>> tuple(tf(T.Image(lucid.rand(3, 32, 32))).data.shape)
(3, 16, 16)
The aspect ratio is kept, so the shortest side lands on the
limit and the other one is longer.