Gregory Lee

There may be a better way to do this, but here is one idea. It does not avoid the for loop, but does result in calling `find_contours` only on the...

This makes me wonder if we should add 'contour' as an object-based region property so it could be used directly via `regionprops_table`.

I am going to close this issue as there is already a preceding issue related to the remaining issue with the `hessian` filter: #3783

Thanks @larsoner, this seems interesting. I will take a closer look later. Another alternative if you have a GPU compatible with CuPy is the affine map implementation in the experimental...

It looks like reslice calls `scipy.ndimage.affine_transform` without any anti-aliasing prefilter. When downsizing, you could first apply smoothing with `scipy.ndimage.gaussian_filter` as in scikit-image's `resize` (see: https://github.com/scikit-image/scikit-image/blob/b7d0209c2b60edfe5ed1b578441c74daab4652d7/skimage/transform/_warps.py#L177-L178 and the lines above for...

> So, will sigma be equal to (1.5/1.25 -1)/2 as per this? That should be a reasonable starting point, but I do not know exactly how that specific value was...

> Thanks a lot for the reply, this sigma was chosen from the link as provided: https://github.com/scikit-image/scikit-image/blob/b7d0209c2b60edfe5ed1b578441c74daab4652d7/skimage/transform/_warps.py#L177-L178 Thanks, I understood that. I meant that I wasn't sure what reference scikit-image...

> In fact, I went looking for it today in cupyx.scipy.ndimage, and almost assumed if it wasn't there it didn't exist. Luckily found https://github.com/rapidsai/cucim/pull/318 and will try it out from...

@jni or @mkcor, this seems ready to go after a 2nd review

Thank you for the nice suggestion @luigifcruz! I will bring it up with the team for possible inclusion in a future release.