Explaining what an image ML model is looking at

[Dale Lane]

I've been working on how students could make a demo that explains what parts of an image are most significant to an image classifier that they train. 

If you'd like to see or try it, it's at https://dalelane.co.uk/blog/?p=4447 

I'm not sure how to turn it into a worksheet yet, mainly as I can't think of a way to make it interesting or more accessible. If anyone would like to help or contribute ideas, I'd be very grateful!