Reasonable theory (capacity) although I just checked today’s notifications/event videos on two of my V2 cameras. On one camera, I’ve had 5 events and on the other I’ve had 4. On each, only 1 detection has been a person detection. In fact, both were correct, the other 7 were detected (properly) as simple motion alerts, sooo … working properly again today so far.
If Carver’s theory (I mean master’s thesis ) is right I kinda wish they’d just cut it out and revert to last week’s rule set, call PD finished, and take no more submitted videos. It was quite good then.
This feature is obviously still very much in beta, I selected the option of paying $0 for an option that was advertised when I purchased the cam. I may end up paying for the full cloud option at some point but for now I have 1 cam with an SD card continually recording. It seems Wyze is going to move away from continual recording (outdoor doesn’t do it for power reasons, doorbell doesn’t have an SD slot for some unknown reason). Maybe then I’ll go with a cloud option, or switch over to a fully local recording system
Aren’t we the people who helped you with testing Person detection? Didn’t you ask us to react if camera cut it right? Honestly, having a lot of trouble with your cameras that I have at home. And I am not going to pay for this.
incorrect , person detection was available for everyone, it was only limited to beta testers for very a short time during the beta testing
—EDIT: I want to clarify upfront that it turns out @HDRock was indeed correct, and I that I simply misunderstood that “Beta” was actually available to everyone. I mistakenly assumed that that the whole time it was in “Beta” it was only available to beta testers and that HDRock was talking about XNOR, not Wyze developed PD…thus making the following post incorrect, misassumptive, and wrong. My apologies to @HDRock for my misunderstanding. This post can be skipped…I only leave it up because I dislike deleting things and confusing context of other posts for others later.
That seems needlessly contentious since it takes the rest of my post out of context if you read it in full…I know it was long…but after the implication that this may decredit my relevant thoughts I will clarify again here:
Yes, Person Detection was first available to everyone, but I was talking about the Cloud PD using cloud AI that Wyze created and needed trained that was recently only available to beta testers for over the last half a year (hardly a short time…in fact XNOR was only available for a very short time, then it switched to Wyze developing their own, which has been limited to beta access only for what I believe is longer than it was officially available to anyone as XNOR (not counting people holding onto the old firmware after the time it was no longer “allowed” by XNOR/Apple and shouldn’t count as approved/official anymore).
Regardless of any of that, since my entire post was specifically about training the Wyze developed cloud based AI (not XNOR which needed no training from people and is unrelated to the Wyze cloud PD I was discussing), that whole XNOR time is not related to my post or thoughts as to why things could have suddenly changed now that Wyze Cloud AI is opening to everyone and allowing anyone to submit videos.
Regardless, what you said is good clarification for new people to be aware that there was an open version nearly a year ago, then it switched to Beta only for a long time, began to work well, then suddenly started having problems after it opened up to a larger group of people. That’s what we were talking about in this limited part of why it may have changed from working well during the time when it was beta users only to getting worse after it was opened to a wider audience.
The cloud beta detection was only limited to Beta testers for a short amount of time, it was then put in the production app but still called Person Detection (Beta). This has caused a lot of confusion, but as @HDRock stated person detection (the WYZE cloud beta) was available to everyone through the production app.
This is when it was rolled out to everyone April 21, 2020
Is there an option for those who only want person detection without the full video and cloud storage?
The video has to be uploaded to the cloud to even see if there is a person in it, so avoiding the cloud would not work
If by “full video” you mean the duration of motion then yeah, the free/NYP PD only stores 12 seconds, versus the other CamPlus service option which would store the whole video.
wow, so IFTT and now Wyze. Who else will follow. If you buy a product which promises something how can you later decide to charge for it?? On top of that the person detection doesn’t really work, I get messages that person detected and there is nothing because they are either gone already or something else moved.
I was not offered the NYP PD but that is exactly what I want
Ah, then I stand corrected, I actually do appreciate the clarification. I was under the impression that when it said “beta” it was still only available to beta testers, and since I was always “beta” I was unaware it was available to the masses. My sincerest apologies for the misunderstanding @HDRock. Thank you for pointing this out (I will place an edit at the top of my post that you were indeed correct).
If you had an account before 11/2019 then you’re entitled whether or not you got the e-mail notice. You might have to contact customer service.
If you don’t qualify then consider either getting CamPlus on a single camera or switching to a brand with local person / human / AI detection. There are a lot of them; you’ll pretty much trip over one if you look at competitive cameras on Amazon.
Me too. Silly me, thinking beta meant beta. It’s ridiculous that WYZE tested beta level options on people using production apps. These people didn’t sign up to test software.
They still had to actually turn it on in the app, so by turning it on that was their ‘signing up’ I am guessing
Sorry dude, I wasn’t trying to discredit you , just trying to have accurate facts
I know that now it was totally my mistake.
Like I said, I misunderstood and misinterpreted. Even then, I wish other people would handle their misunderstandings more similarly to how I ultimately did, including accepting responsibility and editing a clarification later like I did. I’m glad you mods edit or delete all the name calling and emotional outbursts others sometimes do. It really is appreciated! At least I didn’t do anything that embarrassing LoL. Anyway thanks.
Thanks for making me feel better knowing I wasn’t the only one who misunderstood.
We know XNOR.ai ran on the camera. We know this is possible. It was a bad idea to do it in the cloud, and now is biting us customers in the ass.
This should be a bandaid while you put person detection back on the camera, for wired cameras. The Outdoor would use too much power.
Figure out ML and run Person Detection on-device and be done with this yet-another-damned-subscription service. You made the bad decision, go fix it so we don’t have to pay for it.
Also, stop using AWS or GWS or Azure. Way too much $$$. Get a bunch of servers from OVH. Way cheaper. $100 per month vs AWS EC2 equivalents at $500+ per month, even with a 1-3 year commit. Use AWS for spikes only.