Switch branches/tags
Nothing to show
Find file
Fetching contributors…
Cannot retrieve contributors at this time
107 lines (75 sloc) 5.35 KB


Here is a short example demonstrating how you can use this client.

Lets say that we want to create our own private namespace and train it to recognize Guido Van Rossum.

Here are the images which we will use for training our namespace index:

And here is the image which hopefully, after training our index will be recognized as "Guido Van Rossum":

  1. First we create our private namespace named testns (this can be done on the page)

  2. Now we import the module and instantiate the class with our api_key and api_secret (you can get them by registering your application on page):

    >> from face_client import FaceClient
    >> client = FaceClient('API_KEY', 'API_SECRET')
  3. Before training our namespace index I just want to show you that the image is not already recognized:

    >> client.faces_recognize('guido', '', namespace = 'testns')
    {u'no_training_set': [u'guido@testns'],
    u'photos': [{u'height': 375,
                         ...omitted for clarity...
                         u'tid': u'TEMP_F@51b67ae268617da2c99c69091ab8f3b0_cf224a584e806722a7fa15a936ed1367_48.00_41.82_0',
                         u'uids': [],
                         u'width': 47,
                         u'yaw': 31.649999999999999}],
                  u'url': u'',
                  u'width': 500}],
                         ...omitted for clarity...

    As you can see, the "uids" list is empty, meaning that Guido Van Rossum is not yet recognized in our testns namespace.

  4. Saving the tags and training our index

    For saving the tags, we need to provide the tags_save method with the tag ids, which we can obtain by using the faces_detect or faces_recognize method.

    In this example, I will use faces_detect:

    >> response = client.faces_detect(',,')
    >> tids = [photo['tags'][0]['tid'] for photo in response['photos']]
    >> tids

    We can also check that the tags were saved by using the tags_get method:

    >> client.tags_get('guido@testns')
    {u'message': u'Tags saved with uid: guido@testns ,label: Guido Van Rossum',
     u'saved_tags': [{u'detected_tid': u'TEMP_F@cc96b0429a7946711de5693c5ff67c46_cf224a584e80672ea7fa15a936ed1367_47.00_27.83_0',
                  u'tid': u'21319_cf224a584e80672ea7fa15a936ed1367'},
                 {u'detected_tid': u'TEMP_F@e2ee88f20076bc1a60c3629281f34197_cf224a584e80672ea7fa15a936ed1367_48.00_34.93_0',
                  u'tid': u'21321_cf224a584e80672ea7fa15a936ed1367'},
                 {u'detected_tid': u'TEMP_F@33c91a546bbba775628e7d7ca969f7ce_cf224a584e80672ea7fa15a936ed1367_48.35_26.40_0',
                 u'tid': u'21323_cf224a584e80672ea7fa15a936ed1367'}],
     u'status': u'success'}
  5. Now when we have the temporary tag ids, we can use them save to save the tags and train our namespace index:

    >> client.tags_save(tids = ',' . join(tids), uid = 'guido@testns', label = 'Guido Van Rossum')
    >> client.faces_train('guido@testns')
    {u'status': u'success',
    u'unchanged': [{u'last_trained': 1274462404,
                 u'training_in_progress': False,
                 u'training_set_size': 3,
                 u'uid': u'guido@testns'}]}
  6. Now after we have trained our index, lets check if Guido is recognized:

    >> client.faces_recognize('all', '', namespace = 'testns')
    {u'photos': [{u'height': 375,
              u'pid': u'F@2981c22e78cc0f12276825aa0b05df86_cf224a584e80672ea7fa15a936ed1367',
                         ...omitted for clarity...
                         u'roll': -1.3400000000000001,
                         u'tagger_id': None,
                         u'threshold': 60,
                         u'tid': u'TEMP_F@2981c22e78cc0f12276825aa0b05df86_cf224a584e80672ea7fa15a936ed1367_51.00_35.20_2',
                         u'uids': [{u'confidence': 20,
                                    u'uid': u'guido@testns'}],
                         u'width': 18.600000000000001,
                         u'yaw': 36}],
              u'url': u'',
              u'width': 500}],
    u'status': u'success',
                         ...omitted for clarity...

    As you can see by looking at the uids key, Guido was now recognized with a 20% confidence!

For more information about the API and how to use it with Facebook and Twitter, visit the official documentation.