I just got the latest beta and enabled person detection, and so far the results haven’t been great. It’s missed some rather obvious events with kids running around, and tagged several vehicles as “person” events. To be fair, it’s only been one day, and on one camera, hopefully it’ll improve over time.
Also, I don’t see the option to submit events like before, will this be possible in the future?
MODERATOR NOTE: This topic has been moved from the Beta category to #ask-the-community. The Wyze AI Person Detection is included in the public 2.9 app release. However, the feature itself is still in a form of beta testing. Users are gradually being added as server capacity is increased. You will not see the feature as available yet in 2.9 unless your account has been whitelisted. To get on the waitlist to join Wyze AI Person Detection, click this link.
@WyzeShawn No problem, I’ve sent a few to you. Most are cars traveling in front of a driveway, the last one I sent is a squirrel hopping around. Definitely cute, but not a person by most people’s standards.
A lot of times detects cars moving and at night with cars headlights shining onto the street as people. I submitted a bunch of videos through the app. IMO, it’s working better than xnor’s.
Has Wyze considered saving time and partnering with someone like https://www.xailient.com/ ? Their implementation seems extremely similar to what XNOR.AI was (before being purchased by Apple). They have code that is made to run on on low resource devices like these cameras. Just curious why Wyze would reinvent the wheel. Maybe someone at Wyze should balance out which is more feasible for the company. My $0.02…
control. look at the negativity people expressed when xnor.ai backed out and Wyze lost person detection. if they do the person detection in house they have more freedom with how its used, when its used, what it detects and they don’t have that abandonment issue again if another company were to back out of a contract.
im sure once Wyze gets the full sample sizes they need they will start tweaking the code and make it even better than it was before and they will have the added benefit of it being solely theirs.
Personally I support them developing their own code. Xnor was not that accurate and any support needed got filtered thru layers. I think with their own code they will wind up with a better product and be able to give better support and improvement in the long run.
They also have the advantage of an open attitude with their customers which gives them a huge and willing testing base.
May be not compatible with the architecture, or since theyve been burned already may choose to remain inhouse, or possibly being investigated as an avenue but we just arent privy to those details
