So when reviewing a recorded event there is the thumbs up/down buttons. I understand reporting an incorrect tagging with the thumbs down, but with the thumbs up it still takes me to the page to pick what was in the video. Am I still supposed to report with the tag it got correct? Seems excessive and you’d think just hitting the thumbs up would be enough.
My mailbox Cam v3 will occasionally give me a False Person detection at night. I will correct it with a Thumbs Down.
There is one guy that walks through my neighborhood regularly in the middle of the night which if correctly Tagged, I will submit a Thumbs Up. This Event was Tagged correctly.
You might try focusing only on one camera and one Detection Tag (like maybe Person) to do Thumbs Up and Thumbs Down. I feel I am helping the Wyze community and it does not feel overwhelming.
Yet I’ve been submitting incorrect video tags for years and the AI still makes the same mistakes, it’s like it’s not learning at all.
But regarding the thumbs up with a correct tag, am I to understand that after doing the thumbs up I pick the tag it got correct to submit? To me, it seems they should have made it that all we would have to do is click the thumbs up if it’s correct and that’s it. And if the tag was incorrect then that’s where we would also pick what the tag should have been, then submit it.
The other gray area is what if Tagging is only partially correct. In this sample video if Person was Tagged but Not Pet, would do a Thumbs Up or Thumbs Down to add Pet?
I always do a thumbs down if it doesn’t tag both the dog and person.
I don’t know if this helps the discussion here or not, but with Cam Unlimited Pro and the Descriptive Alerts it generates for notifications, their system goes beyond tagging things like “Person” or “Pet” and tries to describe the scene. When I’m reviewing these events that have descriptive text within Wyze Web Portal, I’m able to give it and also choose an option like “False information.” from a list and then type detail about the error into a text box and submit that. I’m not sure how that information is incorporated into the “training”, but it allows me to describe a correct interpretation of the event and point out specifically what’s wrong about what the system told me.
Since I don’t have Cam Unlimited Pro, none of that applies to what I was talking about.
