Having been using my new Wyze V3 in an outdoor environment for a couple of weeks now, I’ve determined that Wyze’s AI is not very intelligent. So far I have found more than 1000 events in my event list which I would think should not even be triggering any sort of motion detection - flashes of lightning, lights turning on and off, shadows from a moving flag, small bits of dust passing by the camera lens, particularly at night. These clips are of course being sent to Wyze’s cloud servers for classification, then being stored as motion events on my camera, events that serve only to clutter up my events list. By my estimation, Wyze has a LOT of work to do before they can truly call their AI intelligent.
There are two different pieces to Wyze motion detection. First, the camera will detect motion, defined as changes in pixels from the image it captures. Wind, dust, insects, changes in lighting. Once captured the Wyze servers process the recording to identify objects such as people vehicles and pets.
The cameras, mostly, do not use PIR (passive infrared) to detect objects so any motion triggers recording. The AI is only used to notify you of specific items of interest. The motion event is always recorded, but you can set notifications for just the AI events.
3 Likes
Those are all reasons that the pixel detection senses “motion”. Some things you can do to lessen or eliminate these is to tighten down your detection zone to only important areas that are less effected by light changes, adjust your sensitivity levels, or adjust your camera mount and viewing angles.
You mention Wyze AI, but didn’t mention how it affects your events? Motion detection on the camera is not the AI, Wyze AI is in the cloud and analyzes your video once the clips are uploaded. It then tags the video with “tags” of the objects in the video that it detects. You see these tags in the event tab next to the clips.
Do you have cam plus?
2 Likes
Yes, the AI doesn’t have anything to do with your events list. All motion that fits the parameters you have set for the camera is sent to the cloud, where the AI evaluates it. If it is recognized by the AI, you get a specialized notification. If not, it is left in the events list in the cloud for you to review, as it did meet your parameters for an event. Nothing is ever deleted. The AI just tags things it recognizes.
You might consider using the detection zone or reducing sensitivity if you are getting more events than you want. The camera will create an event of a person close to the camera even with the sensitivity set to 1. Also, if it is a really busy area, know that even if you only have 1 box active on the detection zone, so long as the person or object passes thru it, an event will be created, and the whole frame will be recorded by the camera and evaluated by the AI.
Also, the events are not being recorded on your camera. They are cloud events listed in the app. You don’t have to delete anything; you can simply let them expire, and the cloud server will automatically delete them in 14 days. 
1 Like
The v3 cameras are pretty much useless without paying the extra money for camera plus. Sure feels a bit like extortion.
Yeah, but if you have your cam set up to record everything to your SD card, at least you can go to the time of the 12 second recording and see as much before and after the 12 second event as you like. Certainly inconvenient but no extra cost.
I have 24 cameras and I am on the $99/yr. CamPlus Unlimited plan that allows up to 99 cameras.
With only my 24 cameras that is less than 35 cents per camera / per month
@murph007 So that is Cheap “Extortion” as you call it.
To me it is a Great Service at a very reasonable price.
Not sure who’s post you are reading, but I never called it extortion. And I would agree that for someone with such extensive use of the service as yourself, you are certainly getting your money’s worth.
1 Like
Sorry about that, it was the post above yours.
Thought I clicked Reply to @murph007
EDIT: I actually did Reply to murph007 it just places posts in chronological order 