How are people differentiated from pets? Is it by size, number of legs, shape? I’m wondering what setting I should use to ensure that I capture deer, but not headlights or leaves, etc.
Is it likely that person detection would pick up a deer? Unfortunately it’s difficult for me to test since not triggering is inconclusive.
Your first question isn’t entirely answerable. Machine Learning (AI) works by people feeding it examples and telling it when it guesses right and wrong until it decides on its own what counts as that object. The more data it has to work with, the more accurate it becomes…but researchers have found that it often makes these decisions VERY differently from how we would assume.
In short, it’s very probable that nobody knows for sure what exact criteria it is using to differentiate between different objects, at least that’s a common thing found among machine learning in general. We can GUESS what it is considering, and there may be some manually added parameters to help, but in the end, a lot of things are unknowable with this field. That’s just how AI/Machine Learning works (we humans don’t program every detail, we humans just feed it data and tell it if the result it guessed was correct or not and it makes it’s own adjustments based on our feedback). The research on it is very fascinating…at least for nerds like me. But back to the answer:
Your best option is to select “Pet” for now because that is the label most people choose for “Any Animal” at this moment. It’s possible that Wyze will eventually make a separate label for wild animals (as they recently did for insects), but at least for right now, most people submit Deer under the “Pet” label, so that is the detection label you are most likely to be notified about deer with.
Detection works best when the object is close and in good lighting. The farther away and the darker the conditions, the less precise/accurate the detections.
It will pick up a deer, Also Bears and skunks, Identifies them all as “Pets” Don’t know about cattle. Birds flying into view are “people” if they are high enough off the ground and “motion” if they are walking/hopping, Tractors are “vehicles” or “sound” depending on the range.
Frankly the AI is a long way from ready for prime time.