The Cam Plus sub is not related to the old PD at all. The $1.99 Cam+ wouldn’t offset the minimum cost of PD, I guess.
The answer to that is YES, regardless of anything else. (Assuming you are pre 11/2019.)
This is truly annoying. How does this fall into good business practices? I bought the Wyze cameras because of the motion detection and as of writing this post I am getting zero alerts. Now I find out I have to pay extra for a feature I already had? Oh wait, so I can name my price? Why interrupt the features from your great product to still give us the option to continue this business transaction as agreed upon to begin with? Stop messing with your “friends” Wyze. Or just be professional and stop making things difficult for your customers. This is such odd behaviour and has the signs of death by a million cuts.
You say “motion detection”. I don’t think motion detection is in question here. It is the “person” detection for which Wyze is asking to pay subscription.
You are correct. I was just pointing out the software has gotten worse. So to clarify. No Person detection or even motion detection. I am a sad “friend”
Maybe we all need to update to a different cámara brand
Um, *welcome to the community"?
Hi Wize.
I would like to support WYZE (person detection) and subscribe at a monthly fee. However WYZE payment page do not accept non US/Canadian credit cards. Will all Eupean users now loose person detection? What solution do you offer non US/Canadian users.
The reason I don’t want to pay for person detection is on my 30 day trial period I never got any alerts even when my neighbor was cutting my grass why pay for something when it never alerts me when someone is there
Look at your ADVANCED settings with the * at the top right.
That is where you set the notification settings for motion and sound etc.
Try using PayPal. Some have reported success that way.
I bought 4 cams because of the person detection was built in, then it got taken away and replaced by an inferior person detection version… now you’re trying to make me pay a monthly subscription per cam to keep the feature that was promised in the box. NO
I have been praising the person detection accuracy. But this week for the first time I had both a false negative - yesterday a package was delivered to my door with nary a peep from Wyze - and a false positive - today I got a person alert for no reason.
Ironically this was days after I took down a flag that had been causing motion alerts all day long for weeks - and yet during that time Wyze never missed a person alert, even though I am not on CamPlus. Go figure.
So I figured I’d be fun sport and play along with Wyze. Signed up for Person Detection yesterday. So far I have received 0 alerts. I even went outside and jumped around in front of my camera with still no notifications.
- Sad Friend
Exactly my situation. For months (literally) the person detection on my V2 cameras (Pilot person detection) has been nearly flawless. Yesterday (9/24) for some reason I had multiple false positives (cars detected as people). I’m not sure that had EVER happened before. Will see if it happens again today or was just an aberration. Note that I haven’t signed up for Cam+ nor have I signed up for Name Your Price PD (yet).
It then missed SEVERAL more person detections today! Plain as day clear weather, same comings and goings, NO interfering motion detection and cooldown periods. Like you I haven’t made any changes yet.
Person Detection has stopped working. Edit: Sorry not entirely, I see a few detections in the event log. Maybe it’s the Alexa announcements that are down? I don’t have time to test right now.
The person detection on my V2s and Pan cams (Pilot person detection) has been nearly flawless for me also.
For me eversince the pay if you like came from Wyze I’ve been getting some person detection with no one there.
Person detected works but not well on all my cams.
Exicution was delayed or too late… person was there but they left slready.
I just signed up for $0 person detection to keep the feature but untill Wyze improves and stabilizes.
If so… I’ll start with a dollar and update as who knows how it goes ?
Gotta be load related. As people respond to the mailing they are now actually enabling and USING the person detection, and Wyze’s service is now unreliable. It was working great while we were skating by underground.
Wyze, please ramp up capacity and fix your problems. Your service has become worse exactly as you are getting people to finally pay for it.
For TLDR people, here is the summary of my long post: PD changes might be partially due to miscommunication of submissions of what is a useful video.
Perhaps the PD Alert changes aren’t entirely due to the launch of the “pay what you want” scheme, but something else that changed around the same time as well.
For example, it used to be PD was only available to Beta testers. Also, you could just submit your event video for Wyze to use, but you didn’t label it in anyway (yes, there is a person, no there is not, it’s a correct alert, it’s a false alert, etc), but instead a Wyze employee presumably had to tell the AI what type of video it should be considered as (more help in accurately identifying people, or things it should filter out as not being people).
Suddenly with the launch of the Pay what you want Scheme, they not only made PD available to non-beta people, but expanded it to others as well as Cam Plus users. In addition, even video submission has changed. Now when you watch a cloud event video, it asks you: “Is this video useful?”
Regardless of whether you select Yes or No, it then asks you whether you’d like to submit the video to Wyze to help their AI.
Presumably, this change was made to make it easier to feed all those videos into the AI to recognize “Good video” vs “Filter out and ignore” type of videos. If Wyze is getting hundreds, thousands, or millions of such videos submitted every day (I submit some almost every day myself), they don’t have enough time for a person to manually review each video and determine how to label it…I presume this was to help with that process.
There are a couple of potential issues here that could be contributing to the reduced AI processing. Note that this shows up whether the event is a Person Detected video, or a Motion Video, or Sound or Automation. It is subjective to the person whether it was helpful or not (not WHY). We don’t know what information is passed on into the AI. For example, a cat lover could mark that YES this event was helpful and submit it, and if not caught manually by a human, the AI COULD think that it should start labeling cats as persons. Maybe a person wants a notification to know when a spider builds a web across their lens, so they mark yes it was helpful to let them know to go remove that stupid web.
What about Detection Zones? Say I have a detection zone set up to not pay attention to people on the sidewalk passing by my house, or across the street. When I submit videos showing my front yard and mark yes or no, I doubt my detection zone limits are passed on in that even video, because they are only locally on my cam, not marked on the video itself. IF that video happens to have a person in the background on the sidewalk in front of my house or across the street, it now confuses the AI because I have marked yes or no based on my detection zone preferences in mind, while the AI has no idea about any detection zones, and is confused why I am saying yes or no when there are people present or lacking in the background that it has been trained to recognize in general as correct or incorrect detections. Now it has a conflict…It’s been told in some videos by some people that this is a correct detection, and in other videos it is told this is an incorrect detection. What about people in cars, some mark yes, others no. Neither are wrong, but the AI doesn’t have the information to understand WHY both are right because it’s never received the detection zone limits, or what people are thinking. The AI has to make modifications based on guessing and ignorance, and sometimes blatant contradictions.
Furthermore, since people are selecting “Yes or No” on ANY category of video (sound, motion, PD, Automation, etc), the AI could confuse one category of helpful in a specific situation vs helpful in another category, especially since the event category trigger is done locally, except for PD.
People may be marking yes or no incorrectly (or inconsistently)…How should you mark it if an event video shows a human, but was only marked as motion, but not as Person. Was it helpful because it captured the person and thus marked YES as helpful, or do you mark no it wasn’t helpful because it didn’t label it as a person? This can confuse some people and cause problems for the AI.
What about Automation videos? Similarly, I sometimes have automation videos that are labeled as “automation” in the events tab…sometimes they will show as “Automation” but never labeled as “person” and sometimes I will get a video labeled as “person” but never as automation, even when the motion sensor was triggered (ie: if it noticed the person first, then the sensor was triggered afterward), and sometimes I get a video that shows the exact same clip twice. Once as Automation and once as Person.
EXAMPLES:
Same clip labeled once as automation, once as Person:
Example of clips that were both PERSON and AUTOMATION, but were only as labeled as one or the other, either though all clips were both person and automation clips and should have been labeled as both:
The point is that a person in my situation might mark yes or no on any of these clips and the AI doesn’t know WHY I marked yes or no.
- Wyze needs to change their video submission to include more factors, such as WHY a person selected Yes or No
a) Identified a person when it should not have
b) Didn’t identify a person, but should have
c) Not person related, but it was still helpful
d) Wasn’t helpful for some other reason (short answer)
e) Was helpful for some other reason (short answer)
f) and so on.
The other thing to consider is that some passive-aggressive malcontents are purposely marking the opposite response and submitting them to screw with the AI in retribution for Wyze revoking the free service that was promised.
Again, Wyze has never given us enough information to know for sure what the issue is, but the above are plausible and logical possibilities that could be contributing to recent changes in detection accuracy.
Ah yes, load is a very plausible possibility!