Facial-Identification

A convenience wrapper for the Microsoft Cognitive Services API.

Usage:

import iden

This library provides a convient three-function flow to: Set up a group of people with training face data Input an arbitrary photo and identify if that photo belongs to someone in the group – and if so, to which member within the group

Step 1: Initialize a training set from a json file

iden.init_from_file(personGroupId, training_file, key)

The above function initializes a training data set from a json file.

personGroupId: A unique ID to reference a group of people (training set) by. Valid characters include numbers, english letters in lower case, '-' and '_'. The maximum length of personGroupId is 64.

training_file: a json file containing an array of person objects. Each person object has three fields:

name: The name of the person.

userData: User-specified data for any purpose. The maximum length is 1KB.

faces: An array of urls; each url should be a frontal photo of the person in question. No other faces should appear in the photo. Valid image size is from 1KB to 4MB. See the provided file for more/

key: A valid Microsoft Cognitive Services subscription key authenticated for the face API. See https://www.microsoft.com/cognitive-services/en-US/subscriptions to acquire a key.

Step 2: Identify identities of arbitrary photos matched against a group

iden.identify(personGroupId, identification_file, key):

The above function accepts a set of photos and assigns confidence parameters to each photo for the people in the group that the photo most likely represents.

personGroupId: The same unique ID used above.

identification_file: a json file containing an array of string urls. Each url should depict between 1 and 10 faces. Each face within each url will be attempted to be identified as a member within the group. The faces within a photo will be specified using pixel coordinates.

key: The same key used above.

Step 3: Clear groups

iden.delete_group(personGroupId, key):

Deletes the group associated with a personGroupId.

Example call:

import iden
iden.init_from_file('test_group', 'training_data.json', 'XXXXXXXXXXXXXXXXXXXXXXXXXXX')
iden.identify('test_group', 'identification_data.json', 'XXXXXXXXXXXXXXXXXXXXXXXXXXX')
iden.delete_group('test_group', 'XXXXXXXXXXXXXXXXXXXXXXXXXXX')