Nest products and services - binding, alternatives and further development activities

Dear Community,

there are several threads about Nest related products and services (thermostats, cams and smoke alarm systems) here on the forums. Therefore I decided to create a new one, that the current relevant information is collected here.

The Nest API was deprecated / closed. So it seems that the Nest binding won’t work anymore. How should we proceed? What is the current state of Google smart actions approach - how useful is this approach for an openHAB binding rewriting?

Former Discussions here showed, that Google wants to be the smart home hub by its Google Assistant cloud based approach. If Google sticks to that idea and isn’t offering an API for OH, then it’ll be necessary to find alternatives, I guess?

I am using a nest cam and I like it a lot. The service of storing your video data for a few days is indeed too expensive (around 60 bucks per year) but the product works great: there is a motion detection service and so it is possible for us to use the system as a cat alarm: when the cat is waiting in front of the porch door, we will get messaged and are able to let her in. Anyway for context recognition it’ll be great to have camera systems with an open API available - any ideas which systems are available for that usage?

Thanks in advance for your reply.

Best regards,

I believe that is the only approach other than succumbing to the Google walked surveillance garden.

I am a heavy Blue Iris user and I use my own AI using YoloV3. I let Blue Iris detect the motion, but my code will analyze the alert image and detect all 80 classes.

Blue Iris now has a subscription based AI through Smart Sentry, which of course only detects human beings from an image capture. Since 1, I did not want to pay for the subscription, and 2, I want more classifications, I coded my own AI specifically for Blue Iris use.

Once an image is detected and confirmed via preset confidence level, I get a message from on my phone :slight_smile: I am planning to publish my software as a 1 time purchase once I cleaned up the code. FWIW, detection takes 400ms on my i7 blue iris box, and using GTX1060 GPU, it takes approx 5ms. I have also added face detection to detect people I know but this requires training your own Yolo dataset, or using a different approach such as DLIB’s face recognition algorithms (also included in my code).

Sample text message:

In short, if you pay subscription for storage, then it’s up to you to decide. My 8TB disk for Blue Iris is way more than enough for a few months of video storage for 26 4K cameras.

Another approach you can potentially use is Orange Assist with a custom rule