Decompilation via circuit images

[chants]

Is there a good dataset out there containing pictures of many circuits e.g. a carry ripple adder, a Wallace tree multiplier, IEEE floating point operations, barrel shifters, leading zero counters, etc and so forth.

Would it then be possible to train a machine learning model to classify parts of circuits using that, so that to decode a complicated chip or even an FPGA circuit which is rendered as an image in a tool, component by component? It would seem to me like a really powerful way to understand or reverse complex logic circuits.