Smart camera, AI, Tensorflow, etc

I am brand-loyal now to Wyze. The cameras, motion detector, and entry senors have been both awesome and crazy cheap. With that being said, I would honestly pay big money for something new. A very low-power AI camera that uses something like TensorFlow, or some opensource API, to detect very specific things AND report those things to the smart devices. Imagine some scenarios with me and for those reading this, please comment your scenarios too:

  1. I have a Wyze Cam 2 on my bird feeders. I would love to detect birds, squirrels, rats (yes, I have captured and found new homes for 6 of them now!), mice, and humans! Imagine recording ONLY the birds and then sharing that data with the Cornell University bird watching app!! Wow. For science!

  2. I would like to record ONLY the humans who linger around my house. I’m not a “get off my lawn” sort of guy. Kids can play in my yard but we have Amazon package snatchers here. Sometimes, the Zmodo cameras don’t catch them. Sometimes the Wyze cam is recording a car driving by and somehow doesn’t populate the events with a human on my porch. This would take all guesswork out of placing the cameras if I could detect specific things and feed that info to the smart devices.

Alexa says, “There is a human on the porch.”
Alexa says, “There are birds at the bird feeder.”
Alexa says, “Oh no, the rat is back.”
Alexa says, “The letter carrier has arrived.” (If we can detect USPS trucks!!)

  1. Sadly, the device would be expensive. Although people are using all of this on Rasberry PI, it takes some CPU power. You’d have to invest in a 14nm- 10nm Arm quad core or better to do this. It would get hot and require a lot of power if it’s not the ideal CPU. Therefore, I WOULD pay more for this. It could do the job of 2-3 Wyze Cam 2s. “Now how much would you pay! But wait, there’s more…”

  2. Technically, when the cameras are this smart then the recordings get narrowed down very specifically and that 32Gb limit on the SD card becomes less important. You could keep it 32Gb or go big and address 64, 128, or 512Gb!

  3. This would clean up our app event fees because like I say in number 4, videos only record for very specific AI triggers. 10 people enter my yard, then there are only 10 videos, whereas now I have cars and lights triggering my videos and I have about 100 events to search through just to see 10 people.

  4. I’m sorry to add this but honestly, it would be so awesome. It could be nice to have a small POE port so that everything can be wired. As it stands now, my Wyze Cam is only as good as my router. Even an Ethernet port + USB for power is fine. This seems like a lot to ask but if anybody could do it, that would be you all!

Ok, that’s my wishlist. This sort of camera will be a “shut up and take my money” deal. :stuck_out_tongue:

Thank you! I’m ordering some more cameras this week and I have shared them with friends. These devices can make your house and the houses around you SAFER!!


To add to this wishlist, I would be glad to pay for add-on ability to enable access to pre-built AI objects as the examples listed above. Also what I would really like is to have an AI Developer option to:

  1. Upload my own training images for given Object Description. Provide access to “Object Description” under Account / Wyse Services / AI Developer / Object Description .

  2. Assuming option 1 would become available, allowing the option to contrib to the AI extras community for inclusion in pre-built AI objects for all users who pay for AI extra access.

Thank you for your consideration.