Dahua Integration(AI Cameras), Frigate, and Double-Take

So I have some fairly higher end Dahua camera’s 8mp(4k) w/ 1/1.2" sensors, that are AI around my property.

The dahua cams themselves have basically perfect Human/Vehicle detection, with literally 0 false positives in the last 3 months. Where as Frigate w/coral, while good, still gets 1-3 false positives a week, per camera.

I was working through trying to setup double-take with deepstack, based off of frigate’s mqtt events, to be able to tag timelines for facial recognition.

I guess what I am getting at, is I would like to forward detections from the dahua integration’s binary sensors, to frigate, as the camera’s AI has zero false positives. Has anyone been able to figure out how to do this? or publish an mqtt message to trigger frigate externally to completely disable object detection in frigate. But still be able to create a ‘person detection’ from a sensor, to trigger double take?

A few things here.

  1. Would be happy to help solve those false positives. With proper tuning and some filtering they can easily be removed in Frigate. My front doorbell cam has not had any false positives in the last 6 months.
  2. We don’t support that in Frigate, there is no input from outside sensors like that.
  3. There is no need for that though, double-take has an API to support manual uploads.

Thanks for the response, I guess I need to look further at double take’s API, to see if I can do it directly from the camera sensors.