diff --git a/src/inception_model.py b/src/inception_model.py
index fddd3e04788031354daf08c347f6cd0f0ccc6f3c..09e032ca70a8e63d9459464e968d25fac29cd4f1 100644
--- a/src/inception_model.py
+++ b/src/inception_model.py
@@ -1,13 +1,8 @@
__author__ = 'Felix Kleinert, Lukas Leufen'
+__date__ = '2019-10-22'
import keras
-from keras.layers import Input, Dense, Conv2D, MaxPooling2D, AveragePooling2D, ZeroPadding2D, Dropout, Flatten, \
- Concatenate, Reshape, Activation, ReLU
import keras.layers as layers
-from keras.models import Model
-from keras.regularizers import l2
-from keras.optimizers import SGD
-from keras.layers.advanced_activations import LeakyReLU, PReLU, ELU
class InceptionModelBase:
@@ -19,13 +14,7 @@ class InceptionModelBase:
self.number_of_blocks = 0
self.part_of_block = 0
self.act_number = 0
- # conversion between chr and ord:
- # >>> chr(97)
- # 'a'
- # >>> ord('a')
- # 97
- # set to 96 as always add +1 for new part of block
- self.ord_base = 96
+ self.ord_base = 96 # set to 96 as always add +1 for new part of block, chr(97)='a'
def block_part_name(self):
"""
@@ -59,7 +48,7 @@ class InceptionModelBase:
"""
self.part_of_block += 1
self.act_number = 1
- regularizer = kwargs.get('regularizer', l2(0.01))
+ regularizer = kwargs.get('regularizer', keras.regularizers.l2(0.01))
bn_settings = kwargs.get('bn_settings', {})
act_settings = kwargs.get('act_settings', {})
print(f'Inception Block with activation: {activation}')
@@ -67,25 +56,25 @@ class InceptionModelBase:
block_name = f'Block_{self.number_of_blocks}{self.block_part_name()}_{tower_kernel[0]}x{tower_kernel[1]}'
if tower_kernel == (1, 1):
- tower = Conv2D(tower_filter,
- tower_kernel,
- padding='same',
- kernel_regularizer=regularizer,
- name=block_name)(input_x)
+ tower = layers.Conv2D(tower_filter,
+ tower_kernel,
+ padding='same',
+ kernel_regularizer=regularizer,
+ name=block_name)(input_x)
tower = self.act(tower, activation, **act_settings)
else:
- tower = Conv2D(reduction_filter,
- (1, 1),
- padding='same',
- kernel_regularizer=regularizer,
- name=f'Block_{self.number_of_blocks}{self.block_part_name()}_1x1')(input_x)
+ tower = layers.Conv2D(reduction_filter,
+ (1, 1),
+ padding='same',
+ kernel_regularizer=regularizer,
+ name=f'Block_{self.number_of_blocks}{self.block_part_name()}_1x1')(input_x)
tower = self.act(tower, activation, **act_settings)
- tower = Conv2D(tower_filter,
- tower_kernel,
- padding='same',
- kernel_regularizer=regularizer,
- name=block_name)(tower)
+ tower = layers.Conv2D(tower_filter,
+ tower_kernel,
+ padding='same',
+ kernel_regularizer=regularizer,
+ name=block_name)(tower)
if batch_normalisation:
tower = self.batch_normalisation(tower, **bn_settings)
tower = self.act(tower, activation, **act_settings)
@@ -128,13 +117,11 @@ class InceptionModelBase:
block_name = f"Block_{self.number_of_blocks}{self.block_part_name()}_"
if max_pooling:
block_type = "MaxPool"
- pooling = MaxPooling2D
+ pooling = layers.MaxPooling2D
else:
block_type = "AvgPool"
- pooling = AveragePooling2D
+ pooling = layers.AveragePooling2D
tower = pooling(pool_kernel, strides=(1, 1), padding='same', name=block_name+block_type)(input_x)
- # tower = MaxPooling2D(pool_kernel, strides=(1, 1), padding='same', name=block_name)(input_x)
- # tower = AveragePooling2D(pool_kernel, strides=(1, 1), padding='same', name=block_name)(input_x)
# convolution block
tower = Conv2D(tower_filter, (1, 1), padding='same', name=block_name+"1x1")(tower)
@@ -179,18 +166,18 @@ class InceptionModelBase:
block = keras.layers.concatenate(list(tower_build.values()), axis=3)
return block
- @staticmethod
- def flatten_tail(input_x, tail_block):
- input_x = Flatten()(input_x)
- tail = tail_block(input_x)
- return tail
-
if __name__ == '__main__':
print(__name__)
from keras.datasets import cifar10
from keras.utils import np_utils
from keras.layers import Input
+ from keras.layers.advanced_activations import LeakyReLU
+ from keras.optimizers import SGD
+ from keras.layers import Dense, Flatten, Conv2D, MaxPooling2D
+ from keras.models import Model
+
+ # network settings
conv_settings_dict = {'tower_1': {'reduction_filter': 64,
'tower_filter': 64,
'tower_kernel': (3, 3),
@@ -203,8 +190,8 @@ if __name__ == '__main__':
pool_settings_dict = {'pool_kernel': (3, 3),
'tower_filter': 64,
'activation': 'relu'}
- myclass = True
+ # load data
(X_train, y_train), (X_test, y_test) = cifar10.load_data()
X_train = X_train.astype('float32')
X_test = X_test.astype('float32')
@@ -214,36 +201,19 @@ if __name__ == '__main__':
y_test = np_utils.to_categorical(y_test)
input_img = Input(shape=(32, 32, 3))
- if myclass:
- googLeNet = InceptionModelBase()
- output = googLeNet.inception_block(input_img, conv_settings_dict, pool_settings_dict)
- else:
- tower_1 = Conv2D(64, (1, 1), padding='same', activation='relu')(input_img)
- tower_1 = Conv2D(64, (3, 3), padding='same', activation='relu')(tower_1)
-
- tower_2 = Conv2D(64, (1, 1), padding='same', activation='relu')(input_img)
- tower_2 = Conv2D(64, (5, 5), padding='same', activation='relu')(tower_2)
-
- tower_3 = MaxPooling2D((3, 3), strides=(1, 1), padding='same')(input_img)
- tower_3 = Conv2D(64, (1, 1), padding='same', activation='relu')(tower_3)
-
- output = keras.layers.concatenate([tower_1, tower_2, tower_3], axis=3)
-
+ # create inception net
+ inception_net = InceptionModelBase()
+ output = inception_net.inception_block(input_img, conv_settings_dict, pool_settings_dict)
output = Flatten()(output)
output = Dense(10, activation='softmax')(output)
model = Model(inputs=input_img, outputs=output)
print(model.summary())
+ # compile
epochs = 10
lrate = 0.01
decay = lrate/epochs
sgd = SGD(lr=lrate, momentum=0.9, decay=decay, nesterov=False)
-
model.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy'])
-
print(X_train.shape)
keras.utils.plot_model(model, to_file='model.pdf', show_shapes=True, show_layer_names=True)
- # model.fit(X_train, y_train, validation_data=(X_test, y_test), epochs=epochs, batch_size=32)
- #
- # scores = model.evaluate(X_test, y_test, verbose=0)
- # print("Accuracy: %.2f%%" % (scores[1]*100))