Face and person detection with Deepstack - local and free!

I am also having this problem that HACS removed the deepstack integration??
Is it not being supported anymore?

I also had this problem with deep stack. But
moved to Frigate instead. Works great for object detection.

But any solution for face recognition?

Since Home assistant 2023.8 i get this error:
AttributeError: module 'PIL.Image' has no attribute 'ANTIALIAS'

I fixed it by replacing ANTIALIAS with LANCZOS in image_processing.py.

Before: self._image.thumbnail(newsize, Image.ANTIALIAS)
After: self._image.thumbnail(newsize, Image.LANCZOS)

1 Like

This really worked for me. How did you find out?

1 Like

Hi, Iā€™ve installed deepstack and doubletake. Itā€™s picking up faces in doubletake UI and iā€™ve trained some, but i cant get hacs to find the integration?

The repository has been removed.
I think you should still be abler to copy the deepstack_object folder to your Custom Components folder in HA - GitHub - robmarkcole/HASS-Deepstack-object: Home Assistant custom component for using Deepstack object detection

Thanks, Iā€™ve done so. However doubletake creates it own sensors in HA which i didnā€™t see before

Hello everyone,

I have ā€˜Deepstack server cpu addonā€™ running and the following configuration in configuration.yaml

image_processing:
  - platform: deepstack_object
    ip_address: 192.168.123.170
    port: 5000
    # custom_model: mask
    # confidence: 80
    save_file_folder: /config/www
    #save_file_format: png
    save_timestamped_file: True
    #always_save_latest_file: True
    #scale: 0.75
    #roi_x_min: 0.35
    #roi_x_max: 0.8
    #roi_y_min: 0.4
    #roi_y_max: 0.8
    #crop_to_roi: True
    targets:
      - target: person
      - target: vehicle
      #  confidence: 60
      - target: car
      # confidence: 40
    source:
      - entity_id: camera.hal_prive_main

But the status of image_processing.deepstack_object_markt_links_main remains unknown

I think you will need to specify the ā€˜confidenceā€™ parameter for the target you wish to identify.

Thank you very much for your response.

This did not help me, Unfortunately!!

Could there be something else wrong?