def _pyramid_ensemble_adaptation(teacher: Model) -> Tuple[Model, int]:
"""
Adapt pyramid ensemble by changing its output to contain each of its submodels outputs.
:param teacher: the pyramid ensemble.
:return: the selective_learning pyramid ensemble and the submodels number.
"""
# Get each submodel's outputs.
output1 = teacher.get_layer('submodel_strong_output').output
weak_1_output = teacher.get_layer('submodel_weak_1_output').output
weak_2_output = teacher.get_layer('submodel_weak_2_output').output
# Create zeros.
weak_1_zeros = ZerosLike(name='weak_1_zeros')(weak_1_output)
weak_2_zeros = ZerosLike(name='weak_2_zeros')(weak_2_output)
# Append zeros to the model outputs which do not predict all the classes.
output2 = Concatenate(name='submodel_weak_1_output_fixed')(
[weak_1_output, weak_1_zeros])
output3 = Concatenate(name='submodel_weak_2_output_fixed')(
[weak_2_zeros, weak_2_output])
# Add activations to the outputs.
output1 = Activation('softmax', name='softmax1')(output1)
output2 = Activation('softmax', name='softmax2')(output2)
output3 = Activation('softmax', name='softmax3')(output3)
# Stack submodels outputs.
outputs = Stack(
axis=1, name='stacked_submodels_outputs')([output1, output2, output3])
# Create selective_learning teacher.
selective_learning_teacher = Model(teacher.input,
outputs,
name='selective_learning_' +
teacher.name)
return selective_learning_teacher, 3
def _load_saved_instance(cls, config_dict: Dict, model_path: str,
tf_model: keras.Model):
from kashgari import utils
embeddings = []
for embed_info in config_dict['embeddings']:
embed_class = pydoc.locate(
f"{embed_info['module']}.{embed_info['class_name']}")
embedding: Embedding = embed_class._load_saved_instance(
embed_info, model_path, tf_model)
embeddings.append(embedding)
instance = cls(embeddings=embeddings, from_saved_model=True)
embed_model_json_str = json.dumps(config_dict['embed_model'])
instance.embed_model = utils._custom_load_keras_model_from_json(
embed_model_json_str)
# Load Weights from model
for layer in instance.embed_model.layers:
layer.set_weights(tf_model.get_layer(layer.name).get_weights())
return instance
class SSD(BaseSSD):
def __init__(self, aspect_ratios=None, image_size=None):
self.model = None
if aspect_ratios is None:
aspect_ratios = [1.]
self.aspect_ratios = aspect_ratios
self.num_boxes = len(aspect_ratios) + 1 if 1. in aspect_ratios else 0
self.create_head_layers()
self.build(input_shape=(image_size, image_size, 3))
def build(self, input_shape):
input_tensor = Input(shape=input_shape)
# vgg block 1
x = Conv2D(64,
3,
padding='same',
activation='relu',
name='block1_conv1')(input_tensor)
x = Conv2D(64,
3,
padding='same',
activation='relu',
name='block1_conv2')(x)
x = MaxPooling2D(2, 2, padding='same', name='block1_pool')(x)
# vgg block 2
x = Conv2D(128,
3,
padding='same',
activation='relu',
name='block2_conv1')(x)
x = Conv2D(128,
3,
padding='same',
activation='relu',
name='block2_conv2')(x)
x = MaxPooling2D(2, 2, padding='same', name='block2_pool')(x)
# vgg block 3
x = Conv2D(256,
3,
padding='same',
activation='relu',
name='block3_conv1')(x)
x = Conv2D(256,
3,
padding='same',
activation='relu',
name='block3_conv2')(x)
x = Conv2D(256,
3,
padding='same',
activation='relu',
name='block3_conv3')(x)
x = MaxPooling2D(2, 2, padding='same', name='block3_pool')(x)
# vgg block 4
x = Conv2D(512,
3,
padding='same',
activation='relu',
name='block4_conv1')(x)
x = Conv2D(512,
3,
padding='same',
activation='relu',
name='block4_conv2')(x)
conv43 = Conv2D(512,
3,
padding='same',
activation='relu',
name='block4_conv3')(x)
x = MaxPooling2D(2, 2, padding='same', name='block4_pool')(conv43)
# vgg block 5
x = Conv2D(512,
3,
padding='same',
activation='relu',
name='block5_conv1')(x)
x = Conv2D(512,
3,
padding='same',
activation='relu',
name='block5_conv2')(x)
x = Conv2D(512,
3,
padding='same',
activation='relu',
name='block5_conv3')(x)
x = MaxPooling2D(3, 1, padding='same', name='block5_pool')(x)
# vgg fc modified as conv
conv6 = Conv2D(1024,
3,
dilation_rate=6,
activation='relu',
padding='same',
name='conv6')(x)
conv7 = Conv2D(1024,
1,
activation='relu',
padding='same',
name='conv7')(conv6)
# extra feature map layers
extra1 = ConvBn(256, 1, name='extra1_shrink')(conv7)
extra1 = ConvBn(512, 3, strides=2, padding='same',
name='extra1')(extra1)
extra2 = ConvBn(128, 1, name='extra2_shrink')(extra1)
extra2 = ConvBn(256, 3, strides=2, padding='same',
name='extra2')(extra2)
extra3 = ConvBn(128, 1, name='extra3_shrink')(extra2)
extra3 = ConvBn(256, 3, name='extra3')(extra3)
extra4 = ConvBn(128, 1, name='extra4_shrink')(extra3)
extra4 = ConvBn(256, 3, name='extra4')(extra4)
extra5 = ConvBn(128, 1, name='extra5_shrink')(extra4)
extra5 = ConvBn(256, 4, name='extra5')(extra5)
# heads
feature_maps = [conv43, conv7, extra1, extra2, extra3, extra4, extra5]
confs, locs, anchors = [], [], []
for i in range(len(feature_maps)):
map = feature_maps[i]
conf = self.conf_layers[i](map)
loc = self.loc_layers[i](map)
anchor = self.anchor_layers[i](map)
confs.append(conf)
locs.append(loc)
anchors.append(anchor)
confs_reshaped = [Reshape((-1, 1))(conf) for conf in confs]
locs_reshaped = [Reshape((-1, 4))(loc) for loc in locs]
anchors_reshaped = [Reshape((-1, 4))(db) for db in anchors]
conf_concat = Concatenate(axis=1, name='scores')(confs_reshaped)
loc_concat = Concatenate(axis=1, name='offsets')(locs_reshaped)
anchor_concat = Concatenate(axis=1,
name='default_boxes')(anchors_reshaped)
self.model = Model(input_tensor,
[conf_concat, loc_concat, anchor_concat],
name='ssd_vgg16')
def init_vgg16(self):
vgg16 = VGG16(weights='imagenet', include_top=True)
for i, layer in enumerate(self.model.layers):
layer_name = layer.name
if layer_name.startswith('block'):
layer_weights = vgg16.get_layer(name=layer_name).get_weights()
layer.set_weights(layer_weights)
fc1_w, fc1_b = vgg16.get_layer(name='fc1').get_weights()
fc2_w, fc2_b = vgg16.get_layer(name='fc2').get_weights()
conv6_w = np.random.choice(np.reshape(fc1_w, (-1, )),
(3, 3, 512, 1024))
conv6_b = np.random.choice(fc1_b, (1024, ))
conv7_w = np.random.choice(np.reshape(fc2_w, (-1, )),
(1, 1, 1024, 1024))
conv7_b = np.random.choice(fc2_b, (1024, ))
self.model.get_layer(name='conv6').set_weights([conv6_w, conv6_b])
self.model.get_layer(name='conv7').set_weights([conv7_w, conv7_b])