Facial recognition & room presence using Double Take & Frigate

Share your Projects!
645

Hi all

I’m trying to add Double Take/Deepstack to my already working Frigate.
Frigate is on a dedicated container in Proxmox, while Double Take and Deepstack are HA addon

This is my DT config

# Double Take
# Learn more at https://github.com/jakowenko/double-take/#configuration
mqtt:
  host: 192.168.1.190
  username: !secret mqtt_user
  password: !secret mqtt_pwd
  topics:
    matches: double-take/matches
    frigate: frigate/events


frigate:
  url: http://192.168.1.43:5000
  attempts:
    latest: 15
    snapshot: 0
  cameras:
    - cam1
    - cam2
    - cam3
    - cam4

detectors:
  deepstack:
    url: http://192.168.1.190:5001
    

confidence:
  match: 50
  unknown: 30

save:
  matches: true
  unknown: true

purge:
  matches: 48
  unknown: 12

I get this message every minute in DT as popup

Cannot read properties of undefined (reading ‘id’)

What can I check?

Thanks

3 Likes
646

strange, I also got this error

2 Likes
647

@customstudioyandexru and @woody4165 there is a solution. Check out Error messages popping up · Issue #313 · jakowenko/double-take · GitHub. The fix is essentially using a patched repo from Mincka

3 Likes
648

this helped me. There is no more error

1 Like
649

Friends, help me to learn how to recognize shoes on the surface!

650

I see that Double Take can use OpenCV, but I don’t know what OpenCV is or how to use. Any starter guides or readme’s on the benefits of using opencv or what it does for Double Take?

651

Jag tror inte opencv stöds. Enligt hemsida så kan man använda

1 Like
652

Can someone point me out in making an automation
So i have Frigate, Double take with Deepstack trained on me, a door lock and a camera.
I make an automation in Node Red like this

Screenshot 2023-06-05 at 23.00.33
Screenshot 2023-06-05 at 23.00.331352×166 28.1 KB

When sensor.double_take_denis is door then fire door open.
But the thing is that sensor.double_take_denis remembers its previous state until changed, i have only one camera detection so it is always “DOOR”
How do i get it changed to none (like occupancy sensors)

653

Hi Guy’s! Double Take is working fine in mij HA set-up, but I got a question: can this be use to do “Object Detection” instead of “Face Recognition”?

654

Seems like most people use Frigate for object detection.

655

Frigate has a very limited set of objects based on the coco set currently and may actually get MORE limited but more accurate geared toward surveillance with frigate plus. For example face detection is being added to frigate which would be more efficient that triggering double take on a person that may not be facing the camera.

656

Thank you for sharing your amazing work.

657

Is it possible to distinguish between real person or a photo with face?

658

But I guess Face detection will be added to Frigate Plus, right?

I really was hoping to have a reliable face detection method. I have been experimenting with Compreface, DeepStack and Codeproject.ai but all lack in reliable and effective responses. Lot’s and lot’s of false positives.

659

What Frigate plus?

660

Frigate plus seems to be a subscription model from Frigate. Don’t exactly what it is.

1 Like
661

For some reason my HA logbook only shows double_take_frigate_office changed to 1 or 0 but isn’t showing any names. How do I adjust this?

662

Same here. And my DT/Compreface setup never shows anyone as ‘unknown’. It just matches everyone to faces it knows and lowers the confidence.

663

@Eelco_Anneveldt the fix is documented here for the error messages Facial recognition & room presence using Double Take & Frigate - #647 by Townsmcp
Errors are related to DT calling home to check for updates on GitHub.
If you are not showing ‘unknown’ for matches that would be down to your settings. Try changing settings to:

  match:
    # save match images
    save: true
    # include base64 encoded string in api results and mqtt messages
    # options: true, false, box
    base64: false
    # minimum confidence needed to consider a result a match
    confidence: 85
    # hours to keep match images until they are deleted
    purge: 24
    # minimum area in pixels to consider a result a match
    min_area: 2500
    #when using substream image, 3600

  unknown:
    # save unknown images
    save: true
    # include base64 encoded string in api results and mqtt messages
    # options: true, false, box
    base64: false
    # minimum confidence needed before classifying a name as unknown
    confidence: 85
    # hours to keep unknown images until they are deleted
    purge: 8
    # minimum area in pixels to keep an unknown result
    min_area: 64

(Important part is the confidence values)

1 Like
664

Thanks a lot for that config! Still finding my way around Double Take. And trying to prevent my girlfriend finding out my home automation confuses her for my ex :slight_smile:

Can I just copy the yaml config for double take, delete the add-on, install the error-free version and paste the config back in? Or is it more complex than that?