Skip to content

Latest commit

 

History

History
101 lines (82 loc) · 2.06 KB

README.md

File metadata and controls

101 lines (82 loc) · 2.06 KB
Raw

Face Recognize

Blocks for working with facial recognition and images

Get Encoding From File

Find a face encoding from an image file, send encoding and other data as signal

Properties

  • Image Path: Full path to the image file that will be added
  • User ID: Id of the face being added to the database
  • Save Name: Name of the face being added to the database

Dependencies

  • face_recognition
  • pickle
  • base64

Input

  • Any signal to trigger the block to run

Output

  • A signal containing the facial encoding, user id, and name

sample output:

{
 'encoding': 'gANjbnVtcHkuY29yZS5tdWx0aWFycmF5Cl9yZWNvbn...',
 'name': 'Barack',
 'user_id': 'bobama'
}

Capture Frame

Grab a frame of video from a specified camera and send the frame data as a signal.

Dependencies

  • opencv-python
  • pickle
  • base64
  • urllib.request
  • numpy

Input

  • Any signal to trigger a frame being grabbed from the specified camera

Output

  • A signal containing the serialized and stringified video frame

Find Face

Grab a frame of video from a specified camera, find a face encoding within the frame, compare the encoding with encoding of known faces from an input signal, output a signal containing the name of the found face.

Dependencies

  • face_recognition
  • opencv-python
  • pickle
  • base64
  • urllib.request
  • numpy

Input

  • A signal through "unknown" to begin collecting frames from the camera and search for faces.
  • A signal through "known" to add the known face encodings and names to compare found faces against. Expects a 'faces' object which contains a list of objects with attributes 'name', 'user_id', 'id', and 'encoding'.

sample known input:

{
 'faces': [
  {
   'name': 'Barack',
   'user_id': 'bobama',
   'id': '4999011a-8ded-49c4-a927-77a09dcdb578',
   'encoding': 'gANjbnVtcHkuY29yZS5tdWx0aWFycmF5Cl9yZWNvbn...'
  }
 ]
}

Output

  • A signal containing the name of the face identified from the webcam

sample output:

{
 'found': 'Barack'
}