def test_simple_model(self):
inputs = tf.keras.Input(shape=(256,)) # Returns a placeholder tensor
# A layer instance is callable on a tensor, and returns a tensor.
x = tf.keras.layers.Dense(128, activation='relu', name='a')(inputs)
x = tf.keras.layers.Dense(64, activation='relu', name='b')(x)
x = tf.keras.layers.Dense(32, activation='relu', name='c')(x)
x = tf.keras.layers.Dense(16, activation='relu', name='d')(x)
x = tf.keras.layers.Dense(8, activation='relu', name='e')(x)
predictions = tf.keras.layers.Dense(10, activation='softmax')(x)
model = tf.keras.Model(inputs=inputs, outputs=predictions)
new_in = model.get_layer(
name='b').input
new_out = model.get_layer(
name='d').output
new_model = model_util.extract_submodel(
model=model,
inputs=new_in,
outputs=new_out)
batch_size = 3
ones = tf.ones((batch_size, 128))
final_out = new_model(ones)
self.assertAllEqual(final_out.shape, (batch_size, 16))
def get_box_classifier_feature_extractor_model(self, name=None):
"""Returns a model that extracts second stage box classifier features.
This function reconstructs the "second half" of the ResNet v1
network after the part defined in `get_proposal_feature_extractor_model`.
Args:
name: A scope name to construct all variables within.
Returns:
A Keras model that takes proposal_feature_maps:
A 4-D float tensor with shape
[batch_size * self.max_num_proposals, crop_height, crop_width, depth]
representing the feature map cropped to each proposal.
And returns proposal_classifier_features:
A 4-D float tensor with shape
[batch_size * self.max_num_proposals, height, width, depth]
representing box classifier features for each proposal.
"""
if not self.classification_backbone:
self.classification_backbone = self._resnet_v1_base_model(
batchnorm_training=self._train_batch_norm,
conv_hyperparams=None,
weight_decay=self._weight_decay,
classes=None,
weights=None,
include_top=False)
with tf.name_scope(name):
with tf.name_scope('ResnetV1'):
conv4_last_layer = _RESNET_MODEL_CONV4_LAST_LAYERS[
self._resnet_v1_base_model_name]
proposal_feature_maps = self.classification_backbone.get_layer(
name=conv4_last_layer).output
proposal_classifier_features = self.classification_backbone.get_layer(
name='conv5_block3_out').output
keras_model = model_util.extract_submodel(
model=self.classification_backbone,
inputs=proposal_feature_maps,
outputs=proposal_classifier_features)
for variable in keras_model.variables:
self._variable_dict[variable.name[:-2]] = variable
return keras_model
def get_box_classifier_feature_extractor_model(self, name=None):
"""Returns a model that extracts second stage box classifier features.
This function reconstructs the "second half" of the Inception ResNet v2
network after the part defined in `get_proposal_feature_extractor_model`.
Args:
name: A scope name to construct all variables within.
Returns:
A Keras model that takes proposal_feature_maps:
A 4-D float tensor with shape
[batch_size * self.max_num_proposals, crop_height, crop_width, depth]
representing the feature map cropped to each proposal.
And returns proposal_classifier_features:
A 4-D float tensor with shape
[batch_size * self.max_num_proposals, height, width, depth]
representing box classifier features for each proposal.
"""
if not self.classification_backbone:
self.classification_backbone = inception_resnet_v2.inception_resnet_v2(
self._train_batch_norm,
output_stride=self._first_stage_features_stride,
align_feature_maps=True,
weight_decay=self._weight_decay,
weights=None,
include_top=False)
with tf.name_scope(name):
with tf.name_scope('InceptionResnetV2'):
proposal_feature_maps = self.classification_backbone.get_layer(
name='block17_20_ac').output
proposal_classifier_features = self.classification_backbone.get_layer(
name='conv_7b_ac').output
keras_model = model_util.extract_submodel(
model=self.classification_backbone,
inputs=proposal_feature_maps,
outputs=proposal_classifier_features)
for variable in keras_model.variables:
self._variable_dict[variable.name[:-2]] = variable
return keras_model