Description

mx.image.imread is undocumented. Searches only lead to examples using it, but never any actual API reference.
I'm led to believe this is coming from OpenCV. It would be nice if this feature was explained in detail, so users know what is available to them.

Various preprocessing examples use mx.image.imread or PIL.Image and their image handling makes a big difference with inference results, throwing off trying to use the examples. Without detailed docs it is hard to understand what is happening.