I have a camera set up on the porch of our vacation home in a setting so rural that it doubles as a “trail cam”. and I check it each morning to see what critters it might have captured. We get everything from bear to bobcat.
But for some reason, when a cloud passes in front of the sun or the sun comes out from behind a cloud I get a pet detection. Not motion. Pet detection. It can be very frustrating. On a partly cloudy day I can have have literally dozens of false detections by lunch.
I’ve been trying to do my due diligence and report these in the hopes that you can improve your detection algorithms, but what adds to the annoyance is that while reporting an issue on my iPhone offers the opportunity to select “lighting change” as what caused the detection, the web view reporting routine offers no such choice and I am compelled to just click “None of the Above” over and over and over and over again!
It makes me wonder if it’s doing any good, given I don’t even have the chance to actually note what caused the event.
If sunlight changes are triggering events, you probably need to reduce your detection sensitivity. I’ve never seen that on mine, only severe changes like car high beams washing across the area will trigger mine.
Next, try to figure out what in the field of view it is identifying as a pet. When it analyzes the image, it doesn’t care where the motion was, if there is anything in the detection zone matching, it will trigger a detection for that object.
You can also make use of “motion tracking” (the green box) to see if maybe the lighting change trigger is actually due to a reflection off something and not just the overall change in light, might help you find stuff to exclude in your detection zone to avoid that.
Thanks for your suggestions. Reducing the sensitivity (to about 30 percent - usually at 50) caused me to miss recording a flock of wild turkeys sauntering through the yard. I happened to be viewing the camera at the time or I wouldn’t have even known.
It is clearly the shadows cast by the sun that trip the detection. There is a concrete casting of a bear on the porch (see clip below for an example of a false positive, casting is in the lower left corner), but it’s hard to believe that the algorithm thinks it’s an animal. Note that most of the detections involving lighting changes show up as motion, but every once in a while it comes up as a pet. Why just those times? I can try excluding that casting, but it won’t stop the lighting change from causing a detection. At least it won’t show up as a pet if you’re right. I’m dubious it will make a difference.
I use the web viewer (my.wyze.com) and it’s obvious that Wyze does not put the same effort into it that they put into the phone app. The lighting problem is just one example. When you’re reporting an erroneous detection with the phone, that’s a choice. Not so on the PC. I have to select “none of the above”.
And there used to be a way to share and download clips through the web site viewer and they removed that feature. They actually made their crappy software worse! I had to use my phone to grab the clip below…
OK yeah that’s a pretty quick and extreme lighting change, are you in the south or midwest? In New England we rarely see something like that, I think something to do with the sun being a bit further away or maybe just different types of clouds here. What you’re seeing is akin to headlights on my yard at night which always trips my cams.
Use detection zone to exclude that casting (maybe the well pump and log in the corner too, who knows which one it is thinking is an animal), but no it won’t help with the lighting change still triggering events unless you turn off “other motion”. I think it is just because your porch stays quite dark but the yard has a significant change in a short time.
You could also try excluding the borders (trees and shrubs in the distance). The cam makes decisions based on how many pixels change so perhaps reducing that will help? Maybe even reducing some dark spots like the porch roof and maybe even the poles will help?
The OG is actually one of the better ones as not detecting lighting changes, when I had one on my front walkway (very small detection zone) it rarely was triggered by headlights. But now that I have a v4 there, it constantly is. I have an OG in another area that gets washed with headlights and again is a pretty small detection zone with sensitivity 100, and it never triggers on headlights. But then for a while I had an OG watching my front lawn (trying to figure out who was letting their dog constantly go in one spot leaving a huge brown patch) and the detection zone was nearly the whole cam, and that did cause it to trigger on headlights a lot. In that case I had it mounted in a second floor window looking down, so the angle may have come into play too. Your tag says OG so I’m assuming that’s the cam you’re using?
Granted my examples are totally different from yours, just thinking that the smaller detection zones seemed to help in my case with lighting changes. If possible (it may not be) you could also try lowering it down so it isn’t at a downward angle as much?
You can reduce the detections by eliminating “other motion” and use only the smart detections, but you risk missing stuff that isn’t recognized. Basically anything tagged as just generic “motion” would no longer log an event and/or alert you (you can select tag and notify separately).
Note that reporting an inaccurate detection isn’t like reporting spam where it always now moves that sender to your junk folder. It is just part of training an algorithm which needs lots of inputs from lots of different people. So it helps, but it isn’t a quick fix.