util.py 9.91 KB
Newer Older
Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
1
2
3
#!/usr/bin/env python
# vim: set fileencoding=utf-8 :
# @author: Tiago de Freitas Pereira <tiago.pereira@idiap.ch>
4
# @date: Wed 11 May 2016 09:39:36 CEST
Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
5
6
7

import numpy
import tensorflow as tf
8
from tensorflow.python.client import device_lib
Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
9

10

Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
11
12
13
14
15
def compute_euclidean_distance(x, y):
    """
    Computes the euclidean distance between two tensorflow variables
    """

16
    with tf.name_scope('euclidean_distance'):
17
        d = tf.sqrt(tf.reduce_sum(tf.square(tf.subtract(x, y)), 1))
Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
18
19
        return d

20
21

def load_mnist(perc_train=0.9):
22
    numpy.random.seed(0)
Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
23
    import bob.db.mnist
24
    db = bob.db.mnist.Database()
Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
25
26
27
28
29
30
    raw_data = db.data()

    # data  = raw_data[0].astype(numpy.float64)
    data = raw_data[0]
    labels = raw_data[1]

31
32
33
34
35
36
    # Shuffling
    total_samples = data.shape[0]
    indexes = numpy.array(range(total_samples))
    numpy.random.shuffle(indexes)

    # Spliting train and validation
37
    n_train = int(perc_train * indexes.shape[0])
38
39
    n_validation = total_samples - n_train

40
    train_data = data[0:n_train, :].astype("float32") * 0.00390625
41
42
    train_labels = labels[0:n_train]

43
44
    validation_data = data[n_train:n_train +
                           n_validation, :].astype("float32") * 0.00390625
45
    validation_labels = labels[n_train:n_train + n_validation]
46
47

    return train_data, train_labels, validation_data, validation_labels
Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
48
49


50
51
52
def create_mnist_tfrecord(tfrecords_filename, data, labels, n_samples=6000):
    def _bytes_feature(value):
        return tf.train.Feature(bytes_list=tf.train.BytesList(value=[value]))
Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
53

54
55
    def _int64_feature(value):
        return tf.train.Feature(int64_list=tf.train.Int64List(value=[value]))
Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
56

57
    writer = tf.python_io.TFRecordWriter(tfrecords_filename)
Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
58

59
60
61
    for i in range(n_samples):
        img = data[i]
        img_raw = img.tostring()
62
63
64
65
66
        feature = {
            'data': _bytes_feature(img_raw),
            'label': _int64_feature(labels[i]),
            'key': _bytes_feature(b'-')
        }
67

68
69
70
        example = tf.train.Example(features=tf.train.Features(feature=feature))
        writer.write(example.SerializeToString())
    writer.close()
Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
71
72


73
74
def compute_eer(data_train, labels_train, data_validation, labels_validation,
                n_classes):
Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
    import bob.measure
    from scipy.spatial.distance import cosine

    # Creating client models
    models = []
    for i in range(n_classes):
        indexes = labels_train == i
        models.append(numpy.mean(data_train[indexes, :], axis=0))

    # Probing
    positive_scores = numpy.zeros(shape=0)
    negative_scores = numpy.zeros(shape=0)

    for i in range(n_classes):
        # Positive scoring
        indexes = labels_validation == i
        positive_data = data_validation[indexes, :]
92
93
94
95
        p = [
            cosine(models[i], positive_data[j])
            for j in range(positive_data.shape[0])
        ]
Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
96
97
98
99
100
        positive_scores = numpy.hstack((positive_scores, p))

        # negative scoring
        indexes = labels_validation != i
        negative_data = data_validation[indexes, :]
101
102
103
104
        n = [
            cosine(models[i], negative_data[j])
            for j in range(negative_data.shape[0])
        ]
Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
105
106
107
108
109
110
111
112
113
114
115
116
117
        negative_scores = numpy.hstack((negative_scores, n))

    # Computing performance based on EER
    negative_scores = (-1) * negative_scores
    positive_scores = (-1) * positive_scores

    threshold = bob.measure.eer_threshold(negative_scores, positive_scores)
    far, frr = bob.measure.farfrr(negative_scores, positive_scores, threshold)
    eer = (far + frr) / 2.

    return eer


118
119
def compute_accuracy(data_train, labels_train, data_validation,
                     labels_validation, n_classes):
Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
120
121
122
123
124
125
126
127
128
129
130
131
    from scipy.spatial.distance import cosine

    # Creating client models
    models = []
    for i in range(n_classes):
        indexes = labels_train == i
        models.append(numpy.mean(data_train[indexes, :], axis=0))

    # Probing
    tp = 0
    for i in range(data_validation.shape[0]):

132
        d = data_validation[i, :]
Tiago de Freitas Pereira's avatar
Scratch  
Tiago de Freitas Pereira committed
133
134
135
136
137
138
139
140
141
        l = labels_validation[i]

        scores = [cosine(m, d) for m in models]
        predict = numpy.argmax(scores)

        if predict == l:
            tp += 1

    return (float(tp) / data_validation.shape[0]) * 100
142
143


Tiago de Freitas Pereira's avatar
Tiago de Freitas Pereira committed
144
145
146
147
148
def debug_embbeding(image, architecture, embbeding_dim=2, feature_layer="fc3"):
    """
    """
    import tensorflow as tf
    from bob.learn.tensorflow.utils.session import Session
149
150
151

    session = Session.instance(new=False).session
    inference_graph = architecture.compute_graph(
152
153
        architecture.inference_placeholder,
        feature_layer=feature_layer,
154
        training=False)
Tiago de Freitas Pereira's avatar
Tiago de Freitas Pereira committed
155
156
157

    embeddings = numpy.zeros(shape=(image.shape[0], embbeding_dim))
    for i in range(image.shape[0]):
158
        feed_dict = {
159
160
            architecture.inference_placeholder: image[i:i + 1, :, :, :]
        }
161
162
163
        embedding = session.run(
            [tf.nn.l2_normalize(inference_graph, 1, 1e-10)],
            feed_dict=feed_dict)[0]
Tiago de Freitas Pereira's avatar
Tiago de Freitas Pereira committed
164
165
166
167
        embedding = numpy.reshape(embedding, numpy.prod(embedding.shape[1:]))
        embeddings[i] = embedding

    return embeddings
168
169
170


def cdist(A):
171
172
173
174
    """
    Compute a pairwise euclidean distance in the same fashion
    as in scipy.spation.distance.cdist
    """
175
    with tf.variable_scope('Pairwisedistance'):
176
177
        ones_1 = tf.reshape(
            tf.cast(tf.ones_like(A), tf.float32)[:, 0], [1, -1])
178
        p1 = tf.matmul(
179
            tf.expand_dims(tf.reduce_sum(tf.square(A), 1), 1), ones_1)
180

181
182
        ones_2 = tf.reshape(
            tf.cast(tf.ones_like(A), tf.float32)[:, 0], [-1, 1])
183
184
185
186
187
        p2 = tf.transpose(
            tf.matmul(
                tf.reshape(tf.reduce_sum(tf.square(A), 1), shape=[-1, 1]),
                ones_2,
                transpose_b=True))
188
189
190
191

        return tf.sqrt(tf.add(p1, p2) - 2 * tf.matmul(A, A, transpose_b=True))


192
def predict_using_tensors(embedding, labels, num=None):
193
    """
194
195
    Compute the predictions through exhaustive comparisons between
    embeddings using tensors
196
197
    """

198
199
    # Fitting the main diagonal with infs (removing comparisons with the same
    # sample)
200
    inf = tf.cast(tf.ones_like(labels), tf.float32) * numpy.inf
201
202
203
204

    distances = cdist(embedding)
    distances = tf.matrix_set_diag(distances, inf)
    indexes = tf.argmin(distances, axis=1)
205
206
207
208
209
    return [labels[i] for i in tf.unstack(indexes, num=num)]


def compute_embedding_accuracy_tensors(embedding, labels, num=None):
    """
210
211
212
213
214
215
216
217
218
    Compute the accuracy in a closed-set
    
    **Parameters**

    embeddings: `tf.Tensor`
      Set of embeddings

    labels: `tf.Tensor`
      Correspondent labels
219
220
    """

221
222
    # Fitting the main diagonal with infs (removing comparisons with the same
    # sample)
223
    predictions = predict_using_tensors(embedding, labels, num=num)
224
    matching = [
225
        tf.equal(p, l) for p, l in zip(
226
227
            tf.unstack(predictions, num=num), tf.unstack(labels, num=num))
    ]
228

229
    return tf.reduce_sum(tf.cast(matching, tf.uint8)) / len(predictions)
230
231


232
233
def compute_embedding_accuracy(embedding, labels):
    """
234
235
236
237
238
239
240
241
242
    Compute the accuracy in a closed-set
    
    **Parameters**

    embeddings: :any:`numpy.array`
      Set of embeddings

    labels: :any:`numpy.array`
      Correspondent labels
243
244
245
    """

    from scipy.spatial.distance import cdist
246

247
    distances = cdist(embedding, embedding)
248

249
250
    n_samples = embedding.shape[0]

251
252
    # Fitting the main diagonal with infs (removing comparisons with the same
    # sample)
253
    numpy.fill_diagonal(distances, numpy.inf)
254

255
    indexes = distances.argmin(axis=1)
256

257
258
    # Computing the argmin excluding comparisons with the same samples
    # Basically, we are excluding the main diagonal
259
260

    #valid_indexes = distances[distances>0].reshape(n_samples, n_samples-1).argmin(axis=1)
261
262

    # Getting the original positions of the indexes in the 1-axis
263
    #corrected_indexes = [ i if i<j else i+1 for i, j in zip(valid_indexes, range(n_samples))]
264

265
266
267
    matching = [
        labels[i] == labels[j] for i, j in zip(range(n_samples), indexes)
    ]
268
269
    accuracy = sum(matching) / float(n_samples)

270
    return accuracy
Tiago de Freitas Pereira's avatar
Tiago de Freitas Pereira committed
271

272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290

def get_available_gpus():
    """Returns the number of GPU devices that are available.

    Returns
    -------
    [str]
        The names of available GPU devices.
    """
    local_device_protos = device_lib.list_local_devices()
    return [x.name for x in local_device_protos if x.device_type == 'GPU']


def to_channels_last(image):
    """Converts the image to channel_last format. This is the same format as in
    matplotlib, skimage, and etc.

    Parameters
    ----------
291
    image : `tf.Tensor`
292
293
294
295
296
        At least a 3 dimensional image. If the dimension is more than 3, the
        last 3 dimensions are assumed to be [C, H, W].

    Returns
    -------
297
    image : `tf.Tensor`
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
        The image in [..., H, W, C] format.

    Raises
    ------
    ValueError
        If dim of image is less than 3.
    """
    ndim = len(image.shape)
    if ndim < 3:
        raise ValueError("The image needs to be at least 3 dimensional but it "
                         "was {}".format(ndim))
    axis_order = [1, 2, 0]
    shift = ndim - 3
    axis_order = list(range(ndim - 3)) + [n + shift for n in axis_order]
    return tf.transpose(image, axis_order)


def to_channels_first(image):
    """Converts the image to channel_first format. This is the same format as
    in bob.io.image and bob.io.video.

    Parameters
    ----------
321
    image : `tf.Tensor`
322
323
324
325
326
        At least a 3 dimensional image. If the dimension is more than 3, the
        last 3 dimensions are assumed to be [H, W, C].

    Returns
    -------
327
    image : `tf.Tensor`
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
        The image in [..., C, H, W] format.

    Raises
    ------
    ValueError
        If dim of image is less than 3.
    """
    ndim = len(image.shape)
    if ndim < 3:
        raise ValueError("The image needs to be at least 3 dimensional but it "
                         "was {}".format(ndim))
    axis_order = [2, 0, 1]
    shift = ndim - 3
    axis_order = list(range(ndim - 3)) + [n + shift for n in axis_order]
    return tf.transpose(image, axis_order)


to_skimage = to_matplotlib = to_channels_last
to_bob = to_channels_first