After 8 months of tweaking every little setting from motion detection sensitivity, detection zones, smart detection etc i cannot for the life of me figure out the best detection zone for my camera
I have a v3 mounted to my front window over looking my porch and front yard and I live on a busy street like 10-15 cars driving up and down my street every hour .
Well my v3 still motion tags every car that drives by and gets triggered by every headlights near and far no matter how many times I have blocked out the road . It still motion tags the area even in the detection zone .
The reason I don’t want it to motion tag every car headlight is bc I don’t want to have my events feed filled by every car that goes up and down my road .
I will upload screenshots of events that are still triggering my v3 , any help here
Best way I’ve found to get around headlights is to turn off the camera’s motion detection at Sundown; then let a motion sensor trigger the camera after that. You still have to be careful about your motion sensor placement so it doesn’t see the cars on the road, neighbors, mowing their lawns, etc.
My motion sensor is on-line 24/7, but my pixel detection is only on during the daytime.
So, as the others implied, your problem is likely that the hesdlights are causing pixel changes even in your defined masked off zones, and there’s not much you can do other than the suggestions offered - using PIR motion sensors and/or focusing only on person detections.
While it wouldn’t eliminate the pixel changes I wonder what would happen if you tilted the camera down so the road were completely out of frame.
Is that a radio tower in the foreground? Can you mount the camera up the tower a ways so that it is looking down at an angle so the road is out of the camera view? Might work far better that looking straight out towards the road.
Yeah, I’ve used other camera brands and this is a big issue with basically any camera that uses pixel change to determine motion. It is my understanding that Wyze is working on training models to see if their AI can ignore shadow and light differences, but this is something that is difficult for everyone in the industry.
There are good suggestions above, and I used to use them myself, including using a PIR sensor and rules at night to make a change. Eventually I just switched to having more advanced AI detection for my front camera just for convenience, and thankfully Wyze’s cloud and AI subscription is extremely reasonably priced compared to most competitors, so it worked out for me that way.
Hopefully someone comes up with a good local solution that can tell the difference between different lighting changes and filter those out to not count as motion in and of themselves.
This is why I’m very critical of the pixel algorithm and the detection zones that are the boxes
I would love to see a pull and drag like eufy or ring . Which I use those cameras and I’m able to just make a pull and drag zone of my yard and front porch and block out the road better for false triggers
I really would like to see wyze give us the option to use a pull and drag . It would definitely be an improvement to the boxes . Just my opinion though
I totally understand. I also have some eufy cams, so I am familiar with what you are talking about. I do like that those kind of detection zone systems do allow us to make more precise detection zone areas in precise angles and shapes. That is one thing I really love about that style of detection zone. I would love for Wyze to implement something that gives us more flexibility and precision in the future. Come to think of it,
You are making a great point. I will raise this topic to the AI team in tomorrow’s AMA event!
Here is the announcement about it:
Basically tomorrow we have the AI team who is going to be answering any number of questions. You are welcome to attend the event and ask them some questions yourself too. If nobody else asks or they haven’t answered a question about by the time I post, I will plan to ask them something like the following:
Not only that but they are more accurate imo. With wyze , any little pixel change and it’ll trigger the cameras so basically my feed is filled of spider webs , cars headlights , light changes , lightning, rain , wind , bees , insects , trees shadows blowing etc but not detecting people or vehicles
With a pull and drag , it would make them more reliable and also help reduce false positives.
With wyze cameras it also comes down to the placement of the camera and so depending on the placement the boxes will take up too much space or not block out a whole area . With pull and drag it would not matter where the camera is placed , I would just pull and drag a zone for the area i only want covered . For my v3 facing my front yard overlooking my street , I have to basically block out the whole screen to not get it triggered , which ultimately means I’m losing out on person detection bc of the detection zone blocking out the whole screen and the AI just labels it as motion…
And yes I’ve read about , I’m looking forward to it . I’m Rulwiz on discord so we’ve talked quite often and I appreciate your help and input in the chats.
Thank you for asking the question I will be unavailable to ask , bc I work nights so by the time the AMA is going to start I will sleeping.
I look forward to reading up on their answers afterwards . Wyze is going in a good direction , i was a bit of a hater in the beginning but I’ve turned a new leaf and am enjoying their products and message .
I was a bit weary and skeptical but I got the thermostat a month ago and after a few tweaks and messing about , I was able to get it to my liking and it is working well now .
Now don’t get me started on my floodlight that’s a whole nother story…