I get a lot of motion events, so much so that some events with people end up getting missed. I’d like the option to only record events with people detected.
If by people you mean dust particles then you’ll get them every time. I am completely on board to what you’re saying. There’s a long way to go with Wyze motion triggering technology.
You could possibly use PIR sensor , it would just depend on your area you are wanting to record. People detection is one of the newer options under beta, still learning my self, how it fully works at the moment
Using motion sensors is a good suggestion but it does require purchasing more hardware and then maintaining batteries. My own personal preference for now is to just use the cameras for my situation…no batteries; although batteries will not bother everyone since they should only have to be replaced a few times a year per sensor…but not for me right now.
Yes, I agree that the ability to have the OPTION of recording ONLY person detected events is crucial in order to avoid the cool down period from missing a person due to the camera being potentially busy recording other motion events such as the shadows of trees/clouds moving around all day.
I do see one issue with the possibility of doing this currently. it is a nice Idea but the current method for people detection is to have the camera record ( either by its own pixel motion, or by a sensor) and then that video gets sent to the cloud, the clips are then given to the proxy partner that uses AI to differentiate between general motion and a person detected in videos. if there is a person detected that clip is marked in someway in the meta data and then you can filter events by those only containing what the AI has flagged as having people in it. Because of the limits of the hardware I doubt our cameras by themselves currently would have the processing power to support native AI algorithms along with everything else it is doing. I’m no programmer but it seems like that might take some ummph and by most standards our cameras are pretty simple and straight forward.
. so going on that I would say individual person detection and recording on the cameras is still a couple models and quite a time away.
If I remember right the timestamps were removed so it would be a clear photo for the training of the AI, but it still makes sense that the actual processing is done off camera to verify what kind of motion it is detected. That would mean sending every video up then deleting any that were not caused by a person, this would defeat the purpose of having the 5 minute cool down since everything would have to be sent up.
ahh maybe I’m confusing it with something else. Ill correct that post
I was told the other day that the person detection is being done by the camera. Can we get an official declaration on this?
Yes, the processing is being done on the camera with the latest beta firmware. That is why you see “Wyze” and “Xnor.ai” in the bottom left corner of the live stream if you have the logo toggled on for the camera.
If the person detection is on-board the camera, then person-only events sounds doable and would be really great, especially if it means fewer or shorter timeouts.
The only problem is with the way it currently works, I am not a programmer so I do not know how hard a change it would be, but currently it watches for any motion then locally checks the video before uploading to see if there is a person. My assumption would be to get it to work you would tell it to check for a person and if there is no person to delete the video and not upload and if there is a person to upload it. Like I said though I am not sure what that entails as far as changing things.
@Jason21271 - that shouldn’t be too difficult if your assumption on the event line is correct. Once the grunt work of determining person/no person is done the decision to upload or not should be a simple matter of checking a flag set during the person/no person process.
I don’t know where the determination itself is made.
If it truly is that easy, I would love to see it implemented
Me to. But I don’t actually know how it is implemented. Probably much more difficult than that.
We need the option to only record Person detected video only . And only start the 5 minute cool down post a person detection . Last night alone I had over 20 false general motion recordings due to bugs and spider webs in front of several cameras . If they would only record Person Detected it would save them Cloud service usage fees by at least 50%.
I assume they are not going to implement this is the free version . Now they have a “Service” they can make money on to provide these types of options .