Our parents adopted a cat a few years ago now, and they haven't installed a cat flap yet because he'll bring home a mouse, and they don't want him to bring them inside the house. So for now, they've been opening and closing the door everytime (they've setup a little security cam at the door so they know when he comes back), which unfortunately involves opening the door in the middle of the night.

So I've been meaning to try and build a "smart" cat flap that utilizes image recognition to detest if he's got a mouse in his mouth or not.

I had previously done a bit of research last year and gave up on it for some reason. Here's to hoping I get a bit further down the road this time, and have some success lol.

[–] d3Xt3r 7 points 7 months ago* (last edited 7 months ago) (2 children)

An easy "cheat" way to do it would be to use ChatGPT 4, you can submit a picture and ask it a question like "is there a cat in this scene and is it holding a mouse? Respond with either 'true' or 'false'".

Python examples here: https://platform.openai.com/docs/guides/vision

[–] AWOL_muppet 3 points 7 months ago

Nice PoC!

[–] NoRamyunForYou 3 points 7 months ago

Huh. Super interesting! I should give this an initial go, and maybe when I come up with my "own" solution, benchmark them against each other. Wonder if there's a trade off (e.g., chatgpt is more accurate but takes longer due to all the back and forth required, whereas the local solution may not be as accurate but is faster)

It's honestly amazing how quickly all this stuff is evolving and growing. Thank you for that tip 👍😄

load more comments (26 replies)