I would like to train a classifier by either replying to an event, or put on an sdcard training shots of 24 different squirrels, 3 raccoons, deer, flying squirrels, bird sounds, birds.
It would also be nice to reply to events , select a category on an sdcard in a camera and be able to go through and rearrange incorrectly placed events or photos in the card and perform the reclassify over.
I also would like to add thousands of classified pictures of individuals pre classified.
I could see a large company using this for maintenance staff for humans. And zoos to provide scene descriptions of chimp cages.
I’ve done this on rpi many different ways. Usb attached coral device, tensorflow, Google cloud services. It would be nice to throw away my hacks and use good stuff.