So, having 1/3 of the screen only marked as a detection zone (lets imagine the bottom 1/3), why when people is walking on the top 1/3 the person detection is triggering???
When this is going to be solved? It has been years waiting. It is like the smartthings integration, never going to happen?
Thanks,
What pixel changes were detected within the detection zone during this time? Person detection comes after the event is uploaded to the cloud and the ai algorithm detects people objects. Can you upload this event? And a pic of your detection zone?
So, here a picture of the detection zone (sorry is not landscape but that feature is not working anymore).
Also a picture of the event (sorry to don’t post the video itself but sharing that video was not working neither).
The thing is, the picture of the event is at the end of the 12 sec. video. You can see the person is on the up left corner, quite far away of the detection zone. Hence, detection zones are not working. When will it work?
Rgds,
But that’s how it works and has always worked, for better or worse. When motion detection is triggered in the defined zone, the clip is uploaded. When the server analyzes it for persons / other AI, it does NOT know of or respect the zone. It looks at the whole frame. "There was a person visible to the camera when motion was detected* is how to think of it.
Sorry to disagree with you but I remember, probably more than 1 year ago, that I have a detection zone and only persons INSIDE the detection zone triggers the notification. Yes, there were cases that if the person was outside the zone but close from it, you have chances to get the notification but, as far of the zone as the example I sent now, did not trigger it.
But are you seeing my point? The leaves obviously triggered the motion detection. Then the cloud server found a person in the frame. There are many threads about this and it is how it’s always worked.
Enable “motion tagging” for troubleshooting. That will mark with a green box what is being determined as motion in the field of view. Motion will be anything that is pixel change shadows moving light moving actual objects moving. And like I said once motion is detected in the detection zone your clip will be uploaded to the cloud and then analyzed for recognized objects. If any objects are recognized, such as person or vehicles, you will be notified as such based on your notification settings. Also to note, that recognized objects aren’t necessarily the cause of the motion, those are just objects that are recognized in the field of view.
Both of you are right but still the question, when will detection zones work properly??.
If a motion is detected in the detection zone and sent to server to analyze, the analysis should happen only in the defined detection zone. If not, why bother to have a detection zone?
Your explanations clarify, even more, that Wyze needs to look at this, one of this decades!!
Thanks,
I agree with you, if at all possible that zone info should factor in to the detection (at least as an option).
I agree with this issue. I get so many notifications that my Cam Plus subscription is next to useless.
As a side note, I read this article from MIT-IBM. It would be nice if they could use this algorithm to move AI detection to the device. https://news.mit.edu/2021/tiny-machine-learning-design-alleviates-bottleneck-memory-usage-iot-devices-1208
Alternatively, send the detection zone information to the cloud along with the video data. It would save them a ton of CPU cycles if they could only evaluate the relevant portion of the frame.
