I've read in several articles/answer written by CS (grad) students or other people that the advanced knowledge of pure math that's required for certain areas of computer vision is sometimes too high for the CS student themselves, and requires the involvement actual math people.
I do wonder if it's true at all? I don't certainly pose this question from a negative view. As a new computer vision researcher who just switched from pure math, I find the math behind CV interesting, but the questions asked to be of different nature. But can the CV researchers with CS background not answer those math questions?