I understand settings > detection settings > motion detection sensitivity as a threshold for the amount of pixel change before the camera detection motion. Does this setting affect the AI determination? In other words, would the basic motion detection have to occur before the video is sent to Wyze for AI analysis? I would think that it works this way? For example, if I want to be able detect a small cat, I would need to dial up the detection sensitivity so that the cam can decide that there is motion, then send the video to Wyze for AI analysis. Is this correct?
Also, regarding motion zones, it seems to me that although the camera will only detect motion (pixel changes) in the defined zone, AI looks at the whole frame and will notify if it “thinks” there is a person, pet, etc. anywhere in the frame, include the area outside of the defined motion zone. Is this correct?
Yes. Motion has to be detected before a clip is made, and the whole of the clip is what the AI evaluates to determine content.
So detection zones and sensitivity count, as that determines whether a clip is sent to the AI. There is no local determination as to whether there is a person, package, pet, or vehicle in the frame.
And kind of impressive if the poster came to those conclusions on their own. Non-obvious stuff that took a while for the forum hive mind to determine.
Thank you. It occured to me that Wyze’s AI looking at the whole frame regardless of the motion detection zone is a limitation and not a feature. It appears that the motion zone and sensitivity settings are local to the camera. The camera does not communicate the motion zone and/or the Wyze AI can only detect based on the full frame. This is why once the video is sent to AI, detectable object anywhere in the frame will be tagged as such.
You are not the first to mention that.