Train on specific pet

Posted this to “wishlist,” but it probably belongs here, sorry about that.

Train to recognize specific pets, analgous to face recognition.

I would use it to trigger on specific activities of specific pets. If it was a trigger, I would absolutely set up a water spray to spray when the wrong cat comes to the food dispenser that is locked to only one cat, and promise to share the videos…

3 Likes

I have similarly wished I could train it to recognize the difference between my pets just like how facial recognition works.

I think we might be able to do this with the new “Wyze Anything Recognition” they announced that they are working on since we’ll be allowed to train the AI to recognize whatever we want. A lot of questions still remain about the implementation of this and what limitations there will be (do we have to train each individual camera, or will it work account-wide like facial recognition, will we have triggers available, or just notifications, etc. We’ll have to see, but I am optimistic that we’ll be able to train it to recognize the difference between our pets so we can use it for some cool use cases.

3 Likes

Agreed. Object recogition tech certainly can do such things (with caveats that two black cats are going to be hard to distinguish, while little cat vs. big dog is a softball). With a trained classifier in the cloud, I imagine it would be useable on any CamPlus device associated with a user account. Hey, that’s a feature to sell more subscriptions!

Triggers would be wonderful. First step is to get the labeling working, then attach the labels to triggers.

1 Like

I will check with the team on this. But my first reaction to this is, how to tell apart the right lovely one from the unwelcomed one? by color, shape, size or other things?

3 Likes

I would like to offer an answer. My background in responding is programming and using Automated Optical Inspection equipment during the manufacturing of circuit boards for communication, medical, and military uses. I have to teach the software what a good part looks like and what defects look like; then be able to tell the difference in a matter of seconds. Specific face/identity recognition takes much finer imaging resolution and a lot of software processing power that is only available on laptops or desktops. Most software sources for it make platforms that can plug in a lot of different high end cameras. The software makers do not make cameras and the camera makers do not make software. It is unlikely to see this in a small camera like Wyze markets.