Gregory Lee

I think the two issues are: 1.) unless `fmt:off` / `fmt:on` blocks are used, it can make some small test matrices formatted in a less readable way. 2.) since upstream...

> On 2 it might be worth asking scikit-image if they are willing to use black. It did come up in the past in https://github.com/scikit-image/scikit-image/issues/4111 and wasn't received too favorably...

I went ahead and ported the binary morphology operations from `cupyimg` over to CuPy in cupy/cupy#3907. I don't think these will appear in CuPy 8.0, though, as `8.0.0rc1` was already...

You might also be interested in [this comment](https://github.com/scikit-image/scikit-image/issues/4083#issuecomment-667611962) and the few following it in scikit-image/scikit-image#4083 It demonstrates that scikit-image's deconvolution can be accelerated by either 1.) enabling multi-threading for the...

> Note: Image processing part (scikit-image compatible APIs using CuPy) is pure Python code and no platform-dependent as long as CuPy is installed. We may want to provide an option...

It looks like there are a number of CUDA implementations around, but I don't know if there are any with Python wrappers. I agree that it would be useful to...

The acceleration provided here will depend on the number and size of the descriptors. Running scikit-image's SIFT on the small `astronaut` image results in 1234 descriptors each of size 128....

This PR now has that same issue we saw previously where the `distance_transform_edt` import fails even though the files are in this branch. ``` src/cucim/core/operations/morphology/tests/test_distance_transform.py:8: in from cucim.core.operations.morphology import distance_transform_edt...

rerun tests

Thanks @m-albert, this is interesting! I had been hoping to look at this at some point for GPU applications where memory limits can sometimes be an issue, so it is...