Questions about the machine learning algorithms (as long as they're actually about the algorithms and not about specific implementations) are clearly on topic here. In this case the fact that the data is coming from vision will probably be relevant, but may not be central to the question.
Questions about what image manipulations make sense to perform are probably more on topic and more likely to get a good response at Signal Processing (about "the art and science of signal, image and video processing"), which has an active image-processing
tag.
Questions about specific implementations of any of the above may be on topic at either of those sites, but also look at Stack Overflow, Computational Science, and maybe Code Review, depending on what the problem is.