Facial recognition & room presence using Double Take & Frigate

Share your Projects!
773

Hi,

So I succeeded in setting up the Frigate-Compreface-DoubleTake combination. And so far it is going wel.

If a match is found, this is correctly published to the relevant sensor:

image
image1351×465 27.2 KB

I’ve tried to set up a notification based on the template provided in the documentation.

This automation has the following condition

condition:
  - condition: template
    value_template: '{{ trigger.to_state.state != trigger.from_state.state }}'

My MQTT sensor has the camera name in the state field (see screenshot). And that doesn’t change (the various attributes change though). What I notice is that my automation gets triggered, but doesn’t pass the condition check and as a consequence no notification is send.

I could obviously remove that condition, but I guess it’s there for a reason?

774

I don’t really know the answer to your question but there is also highly configurable blueprint you could use instead: https://github.com/SgtBatten/HA_blueprints/tree/main
Give it a try.

775

When I press upload in order to train new images for faces it opens my album and I select one or more but nothing gets uploaded. I’ll try on desktop tonight but is this a known thing on mobile? Been so long since I uploaded new images manually.

776

I can confirm that upload under Android does not work. Using Windows it does.

777

Thanks, desktop worked

778

I created a Bug report here:
https://github.com/skrashevich/double-take/issues/618

779

Hi all,

I have setup up frigate + compreface + double take.

I can see known persons that I have trained. But for some reason the mqtt sensor and the notify template is not working.

Can someone point me in the right direction, I want to trigger a automation when there is a known person detected.

780

@Jeroen_Brouwers
@Jako

hello,

i set everything up and he finds the person that i trained, but the mqtt sensor doesnt work, the sonsor gets no update?!

marc
marc404×542 89 KB

mqtt:
  host: 192.168.0.73
  username: marc
  password: !secret mqtt_password
  client_id: double-take
  topics:
    # mqtt topic for frigate message subscription
    frigate: frigate/events
    #  mqtt topic for home assistant discovery subscription
    homeassistant: homeassistant
    # mqtt topic where matches are published by name
    matches: double-take/matches
    # mqtt topic where matches are published by camera name
    cameras: double-take/cameras


mqtt:
  sensor:
    - name: "marc_frigate"
      icon: mdi:account
      state_topic: 'double-take/matches/marc'
      json_attributes_topic: 'double-take/matches/marc'
      value_template: '{{ value_json.camera }}'
      availability_topic: 'double-take/available'

Screenshot 2025-06-01 141251
Screenshot 2025-06-01 1412511060×666 34.5 KB

781

Hi there, guys! I’m currently using jakowenko/double-take v1.13.11.8 in a Docker Container (installed from skrashevich/double-take:latest). I have it running in a rPi, together with HA, where I have ring-mqtt working, and serving snapshots. I am NOT using frigate (can’t have it in this tiny hw), but I’ve got everything else working. Ring detects motion, HA sees the cameras from MQTT, double-take gets the messages from MQTT & processes them through AWS Rekognition, and finds a match! (here’s a manual update)

25-07-13 22:26:01 info: {
  id: 'ade38c0a-e15b-4efc-84c4-fbab4f5b7b05',
  duration: 3.34,
  timestamp: '2025-07-13T22:26:01.720Z',
  attempts: 1,
  camera: 'camera1',
  zones: [],
  counts: { person: 1, match: 1, miss: 0, unknown: 0 },
  matches: [
    {
      name: 'xxxxxxx',
      confidence: 99.98,
      match: true,
      box: [Object],
      type: 'manual',
      duration: 2.83,
      detector: 'rekognition',
      filename: 'ed23a800-123a-46f7-9eec-661e39549aab.jpg'
    }
  ],
  misses: [],
  unknowns: []
}

However, I don’t get double take to post the results in MQTT for HA to “see”, and I get this error message:

25-07-13 22:26:01 error: TypeError: MQTT: recognize error: Cannot read properties of undefined (reading 'DEVICE_TRACKER_TIMEOUT')
    at /double-take/api/src/util/mqtt.util.js:278:32
    at Array.forEach (<anonymous>)
    at module.exports.recognize (/double-take/api/src/util/mqtt.util.js:239:13)
    at module.exports.start (/double-take/api/src/controllers/recognize.controller.js:184:10)
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)

Is this in any way related to my lack of frigate? (the only timeout parameter I see in the config is for it)?

Here’s my complete config:

detect:
  match:
    save: true
    base64: false
    confidence: 60
    purge: 168
    min_area: 10000

  unknown:
    save: true
    base64: false
    confidence: 40
    purge: 8
    min_area: 0

logs:
  level: info
  #verbose
 
detectors:
  rekognition:
    aws_access_key_id: xxxxxx
    aws_secret_access_key: yyyyyy
    aws_region: us-east-1
    collection_id: double-take
    opencv_face_required: false

mqtt:
  host: localhost
  device_tracker_timeout: 30

  topics:
    frigate: frigate/events
    homeassistant: homeassistant
    matches: double-take/matches
    cameras: double-take/cameras

cameras:
  device_tracker_timeout: 30
  camera1:
    snapshot:
      topic: ring/fce1e764-7caf-4098-a1a7-xxxxxxxx/camera/54e01xxxxxx/snapshot/image
      quality: 100
      height: 1080
      crop: True

any help would be greatly appreciated!! (I’d rather not have to resort to parsing the logs to see if there’s been a match :frowning: )
Thanks!

1 Like