Here are some things I learned:

1. You can download 11 million Smithsonian JSON objects at https://github.com/Smithsonian/OpenAccess, 3 million have images, 2 million of those are herbs. Parse for the idsId key to find those with images.