Great question. I have done research into this specificaly in the context of image recognition and natural language procrssing.
I have found some excellent visuals on the blog of Dr. Satya Mallick. Here is an introductory post from him that links to several further explanations. His stuff deals primarily in methods of ML as they pertain to image processing. As I think about it, maybe that’s why he leans toward sophisticated visuals ;).
Disclaimer on Dr. Mallick: I would not call his instruction materials or his demeanor ‘inclusive’ (so don’t imitate the way he talks, lol). That having been said, whoever designs his diagrams does an excellent job.
He does a course called Learn OpenCV, and the first module of that does a great visual explanation of the convolution kernel.