Merge branch 'master' of gitlab.idiap.ch:bob/bob.pad.voice

bob/pad/voice/algorithm/gmm_algorithm.py

@@ -50,8 +50,8 @@ class GmmAlgorithm (Algorithm):
         INIT_SEED = INIT_SEED,
       # parameters of detection pipeline
         performs_projection = True,
-        requires_projector_training = True,
-        use_projected_features_for_enrollment = True,
+        requires_projector_training = False,
+        use_projected_features_for_enrollment = False,
         multiple_model_scoring = None,
     )
     # copy parameters

bob/pad/voice/extractor/__init__.py

@@ -5,6 +5,7 @@ from .glcms import GLCMs
 from .spectrogram_extended import SpectrogramExtended
 from .lbp_histograms import LBPHistograms
 from .dummy_tensorflow import DummyTF
+from .audio_tensorflow import AudioTFExtractor
 
 # gets sphinx autodoc done right - don't remove it
 def __appropriate__(*args):
@@ -28,5 +29,6 @@ __appropriate__(
     SpectrogramExtended,
     LBPHistograms,
     DummyTF,
+    AudioTFExtractor,
     )
 __all__ = [_ for _ in dir() if not _.startswith('_')]

bob/pad/voice/extractor/audio_tensorflow.py

+#!/usr/bin/env python
+# vim: set fileencoding=utf-8 :
+# Tiago de Freitas Pereira <tiago.pereira@idiap.ch>
+
+"""Features for face recognition"""
+
+import numpy
+import bob.io.base
+from bob.bio.base.extractor import Extractor
+from bob.learn.tensorflow.network import SequenceNetwork
+from bob.learn.tensorflow.utils import Session
+from bob.learn.tensorflow.datashuffler import DiskAudio
+
+import bob.io.base
+import bob.ip.base
+
+import logging
+
+logger = logging.getLogger("bob.pad.voice")
+
+class AudioTFExtractor(Extractor):
+    """
+
+    **Parameters:**
+
+    feature_layer: The layer to be used as features. Possible values are `fc1` or 'fc2'.
+
+    """
+
+    def __init__(
+            self,
+            feature_layer="fc1",
+            **kwargs
+    ):
+
+        Extractor.__init__(self, requires_training=True,  # it's True, so that extractor can be loaded from file
+                           split_training_data_by_client=False,
+                           skip_extractor_training=True, **kwargs)
+
+        # block parameters        
+        import tensorflow as tf
+        self.session = tf.Session()
+
+        # self.session = Session.instance().session
+        self.feature_layer = feature_layer
+
+        self.data_reader = DiskAudio([0], [0])
+
+        self.dnn_model = SequenceNetwork(default_feature_layer=feature_layer)
+
+    def __call__(self, input_data):
+        """
+        """
+        # create empty labels array, since this what read/write function of Base accepts
+        rate = input_data[0]
+        wav_sample = input_data[1]
+
+        logger.debug(" .... Extracting frames on the fly from %d length sample" % wav_sample.shape[0])
+        frames, labels = self.data_reader.extract_frames_from_wav(wav_sample, 0)
+        frames = numpy.asarray(frames)
+        logger.debug(" .... And %d frames are extracted to pass into DNN model" % frames.shape[0])
+        frames = numpy.reshape(frames, (frames.shape[0], -1, 1))
+
+        projection_on_dnn = self.dnn_model(frames, self.feature_layer)
+        return numpy.asarray(projection_on_dnn, dtype=numpy.float64)
+
+    # re-define the train function to get it non-documented
+    def train(*args, **kwargs): raise NotImplementedError("This function is not implemented and should not be called.")
+
+    def load(self, extractor_file):
+        logger.info("Loading pretrained model from {0}".format(extractor_file))
+        self.dnn_model = SequenceNetwork()
+        self.dnn_model.load_hdf5(bob.io.base.HDF5File(extractor_file), shape=[1, 6560, 1])
+        # self.dnn_model.load(extractor_file, clear_devices=True)
+
+        #hdf5 = bob.io.base.HDF5File(extractor_file)
+        #self.lenet.load(hdf5, shape=(1,125,125,3), session=self.session)
+
+
+audiotf = AudioTFExtractor()

develop.cfg

@@ -12,7 +12,6 @@ eggs =  bob.pad.voice
         bob.pad.base
         bob.db.base
         bob.measure
-        bob.ap
         bob.db.asvspoof
         bob.db.asvspoof2017
         bob.db.avspoof
@@ -38,7 +37,6 @@ develop = src/bob.bio.spear
          src/bob.pad.base
          src/bob.bio.base
          src/bob.db.base
-         src/bob.ap
          src/bob.extension
          src/bob.learn.tensorflow
 	     src/bob.bio.db
@@ -58,7 +56,6 @@ bob.db.asvspoof = git git@gitlab.idiap.ch:bob/bob.db.asvspoof.git
 bob.db.base = git git@gitlab.idiap.ch:bob/bob.db.base.git
 bob.bio.base = git git@gitlab.idiap.ch:bob/bob.bio.base.git
 bob.pad.base = git git@gitlab.idiap.ch:bob/bob.pad.base.git
-bob.ap = git git@gitlab.idiap.ch:bob/bob.ap.git
 bob.db.voicepa = git git@gitlab.idiap.ch:bob/bob.db.voicepa.git
 bob.extension = git git@gitlab.idiap.ch:bob/bob.extension.git
 bob.learn.tensorflow = git branch=epoch git@gitlab.idiap.ch:bob/bob.learn.tensorflow.git

setup.py

@@ -142,6 +142,7 @@ setup(
 
         'bob.pad.extractor': [
             'cqcc20e = bob.bio.spear.config.extractor.cqcc20:cqcc20',  # Extractor (reads Matlab files) for CQCC features
+            'audiotf            = bob.pad.voice.extractor.audio_tensorflow:audiotf',  # For audio tensorflow
             'dummytfe            = bob.pad.voice.extractor.dummy_tensorflow:dummytf',  # For tensorflow
             'glcms              =  bob.pad.voice.extractor.glcms:extractor',
             'lbp-hist              =  bob.pad.voice.extractor.lbp_histograms:extractor',
@@ -167,6 +168,10 @@ setup(
 
         ],
 
+        'bob.bio.extractor': [
+            'audiotf            = bob.pad.voice.extractor.audio_tensorflow:audiotf',  # For audio tensorflow
+        ],
+
         'bob.pad.grid': [
             'modest         = bob.bio.spear.config.grid.modest:grid',
         ],