support conversions from/to other frameworks

johnnychen94 commented 2 years ago

We do have colorview/channelview/rawview to build these functionalities, but from a user's perspective, it would be much much easier to understand and use things like:

to_matlab/from_matlab
to_torch/from_torch

Because other framework doesn't always contain colorspace information on their numerical array, assumptions must be made. For instance:

for MATLAB we can assume that 3D numerical array with size (m, n, 3) is an RGB image of size (m, n).
for deep learning frameworks like pytorch, we can assume that 4D tensor with size (w, h, 3) is an RGB image of size (m, n).

These utilities do not need to cover every use case and do not need to provide 100% consistency; they're meant to help users to migrate the old codes from other frameworks.

Jay-sanjay commented 1 year ago

sir, can you precisely tell what actually I need to do to fix this issue

ashwani-rathee commented 1 year ago

Johnny was working on it here: https://github.com/JuliaImages/ImageCore.jl/pull/179

JuliaImages / ImageCore.jl

support conversions from/to other frameworks #178