Anyone else get a new submit feedback dialog? It has a checkbox to choose whether to share the video or not. I’m wondering how that works. If you choose not to submit does it still help the training?
It only showed up once for me.
Anyone else get a new submit feedback dialog? It has a checkbox to choose whether to share the video or not. I’m wondering how that works. If you choose not to submit does it still help the training?
It only showed up once for me.
everything has remained the same for me, but keep in mind in order for it to advance the training it must be submitted. once marked correctly or incorrectly and submitted they are put into a “pool” to train the algorithm. the bigger the sample “pool” the better the computer can distinguish what is classified as person and what isn’t. as important as submitting correctly marked persons is, submitting those mistakenly marked for person and marked as NOT having a person is just as important and helps get rid of false positives. the bigger and more diverse the “pool” is the more accurate and faster updated the AI will be.
there are some videos for instance say you forgot clothing or a towel and had to run around your house for a second after a shower in a condition you wouldn’t want someone else to see. this is the type of video most wouldn’t want to have submitted for training and because of privacy concerns, the reason videos are not just automatically submitted. Wyze understands the need for privacy for some events of owners and this is the reason submission of videos is completely voluntary. only you know best what you want to keep private and what would make a good video for training the computer.
I would say submit anything you feel comfortable with. some people understandably wouldn’t want their kids used for the training. submit all that you can COMFORTABLY. all is greatly appreciated,not only by Wyze, but by the larger community as well.
How to upgrade firmware to that??
In the last few days person detection is working great. Very accurate with no false triggers.
@WyzeShawn any update on when V1 will be supported as you guys mentioned in the announcement?
Been very impressed with person detection over the last week, much improved from onboard analysis.
One thought, now that detection processing has been offloaded will it be possible for both RTSP and person detection? Is the camera already using RTSP to continuously send footage to the detection cloud?
Yes, it should be possible since the RTSP firmware still sends the clips to the cloud.
Problem here is both the V1 & RTSP cams are not whitelisted for whatever reason, and was hoping that @WyzeShawn would chime in on why not, when they said at least the V1 will be supported.
I’ve got to say that the new person detection appears to be better than the old one in some ways. I have to cameras facing out a window facing opposite directions with their views crossed.
With the old Xnor,ai person detection, the camera facing out the half covered by a insect screen would never detect a person. The new person detection doesn’t appear to suffer from this very much. There’s still times when the camera facing out the non-screen half picks up person and the screened half doesn’t, but it’s much improved over the old detection.
Wyze has exceeded my expectations in both performance and bringing person detection back.
I became a beta tester about a week ago and was able to upgrade my cameras (v2) to the 4.9.5.105 firmware but haven’t been able to enable the person detection feature under the Wyze services tab.
Do I need to do something else to make it work?
When you say Beta tester, are speaking or the general beta testing for beta apps and firmware?
or was this specifically the “person detection beta” you got signed up for? some haven’t realized that the latest person detection beta was only to a select number of users and you had to jump through a couple of extra hoops to get signed up on that list.
Have you updated the app?
I signed up for the beta apps and firmware. Would you mind explaining those extra hoops to get signed up on that list?
I believe I’m on the latest (v2.9.20) for iPad.
this is the post from andy at Wyze.
there is a link to be signed up for the PD test cycle although I dont know if they are still accepting testers for it. but it can’t hurt to try. this will also give the majority of information about the roll out and 2.9 in general.
Hello,
I just want you guys to know, I check my camera notifications today. The kids were outside running around all afternoon, and it correctly recognized everyone of them. I am quite impressed. Keep up the good work.
Thanks for the info. It is greatly appreciated.
Really wish they were more transparent on this, I know people that signed up after me, and they got access, yet, I still don’t.
So, what is the criteria they are picking people, or is it all random?
sure of the exact criteria but I know the reason they’re doing it this way is to test load their servers and make sure it doesn’t get overloaded in Crash. Once they know they have the server power to run everything then it will be rolled out in full fashion