Hey all.
I’ve created a few PRs for this integration but am not sure how often @bar checks them.
The biggest one is adding a service to trigger cameras and camera groups.
My primary use case for this is triggering a handful of cameras when my driveway sensor goes off though I have some other cases where I don’t want to rely on motion detection and this addresses that.
If anyone wants to test it out, let me know how it goes.
See release 1.0.10 for setting it.
It can be retrieved also and I am looking into it. Still new to HA programming so haven’t settled on the details but I figure a disabled sensor is the best route. Time to learn how to code a sensor
If you want to submit a feature request in github we can track it there.
I’m at a loss. I’ve added the BlueIris repo from HACS. I’ve rebooted - core, host, all of it - several times. When I go to configuration → integrations and try to add Blue Iris, it doesn’t show up.
I see the integration is recognized in the logs (I see the warning about the untested integration), but no matter what I do, I am unable to add it.
I’m running core 2202.3.8, supervisor 2202.03.5, and HAOS 7.5.
The BI integration from HACS is version 1.0.12
I must be missing something, somewhere, but I can’t figure out what. Can anyone assist?
But, now I see that this is no longer being supported… Before I go shelling out $ for it, and for Deepstack… Any other recommendations for a different solution that runs on Windows 10 and does NVR/facial recognition/object recognition?
More than happy to do what I can. @bar setup a great integration. There’s a definite learning curve and pretty sure I am doing things the hard way at times
@kramttocs is maintain the integration as mention by his post, I’m sure all issue will get resolved as will get more into the details of the integration.
I switched to another system and wrote an integration for it as well - Shinobi Video, it supports deepstack as well (integration that I have added to Shinobi Video)
Both of your efforts are greatly appreciated. I’m (mostly) just a freeloader here that benefits off the hard work you guys have done (unless you happen to have a Yamaha Receiver or a Ubiquiti Dream Machine?). LOL
So, I have BI set up, and cameras added to HA, and I have DS installed and configured all the alerting and JPG stuff to actually perform the object detection. I set it up for facial recognition, but I haven’t trained it yet, still learning how to do that.
What I haven’t found yet is any kind of tutorial on how to get DS notifications into HA so that I can take actions on them. Such as, “if it’s during these hours, and person detection says I’m home, if this person shows up in the driveway, open the garage door for them”. I’ve found vague references to the need to use MQTT, but honestly - beyond spelling that, I have no idea what it is. If there’s any kind of comprehensive how-to that one of you kind folks could point me at, I would be most appreciative.
Haven’t used that much the BI integration with DS, but most likely the approach of MQTT will work,
I’m using NodeRED for complex flows, one of the has similar result - identify a specific face - open the gate,
If you can post an example of the MQTT message it can be helpful
I wish I could post an example MQTT message. I don’t even know how to properly format one, or get this thing to alert on recognized faces. I think I’ve got it trained to recognize a face? I have to go test it out.
At this point, I may be abandoning the whole thing. I cannot get the integrated Blue Iris to reliably detect faces - it can’t even tell the difference between myself and my wife (and she doesn’t have a beard!). In trying to integrate these two products, they have really destroyed DS…
Yeah, it’s frustrating. DS all by itself for object recognition was doing pretty good once I had it tuned. But then trying to get it to recognize faces was an exercise in futility.
There’s a way to train DS to recognize faces by using an HTTP post to an API with Postman so you can have multiple faces associated to a single user. However, BI doesn’t respect that, and instead forces you to do “Jim” and “Jim1” and “Jim2” and “Jim3” individually, and then reference them as “Jim*” in the notifications. Also, since it has to be done through the UI, it’s horribly tedious…