vgg16_from_scratch.py

from keras import backend as K
from keras.models import Sequential, Model
from keras.layers import Dense, Activation, Flatten, Lambda, Conv2D
from keras.layers import GlobalAveragePooling2D, Input, Dropout
from keras.layers.convolutional import MaxPooling2D
from keras.models import Model, load_model
from keras.preprocessing.image import ImageDataGenerator
from keras.optimizers import Adam, RMSprop, SGD
import numpy as np
import urllib


path = '/home/ubuntu/nbs/data/dogscats_redux/'
model_path = path + 'vgg16_weights_tf_dim_ordering_tf_kernels.h5'
target_size=(224, 224)
batch_size=32

def preprocess_image(im):
    vgg_mean = np.array([123.68, 116.779, 103.939], dtype=np.float32)
    #im = cv2.resize(cv2.imread(path), (224, 224)).astype(np.float32)
    im = (im - vgg_mean)
    return im[:, ::-1] # RGB to BGR

def create_vgg16(x=None):
    # we initialize the model
    model = Sequential()

    # Conv Block 1
    model.add(Lambda(preprocess_image, input_shape=(224,224,3), output_shape=(224,224,3)))

    model.add(Conv2D(64, (3, 3), input_shape=(224,224,3), activation='relu', padding='same'))
    model.add(Conv2D(64, (3, 3), activation='relu', padding='same'))
    model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2)))

    # Conv Block 2
    model.add(Conv2D(128, (3, 3), activation='relu', padding='same'))
    model.add(Conv2D(128, (3, 3), activation='relu', padding='same'))
    model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2)))

    # Conv Block 3
    model.add(Conv2D(256, (3, 3), activation='relu', padding='same'))
    model.add(Conv2D(256, (3, 3), activation='relu', padding='same'))
    model.add(Conv2D(256, (3, 3), activation='relu', padding='same'))
    model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2)))

    # Conv Block 4
    model.add(Conv2D(512, (3, 3), activation='relu', padding='same'))
    model.add(Conv2D(512, (3, 3), activation='relu', padding='same'))
    model.add(Conv2D(512, (3, 3), activation='relu', padding='same'))
    model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2)))

    # Conv Block 5
    model.add(Conv2D(512, (3, 3), activation='relu', padding='same'))
    model.add(Conv2D(512, (3, 3), activation='relu', padding='same'))
    model.add(Conv2D(512, (3, 3), activation='relu', padding='same'))
    model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2)))

    # FC layers
    model.add(Flatten())
    model.add(Dense(4096, activation='relu'))
    model.add(Dense(4096, activation='relu'))
    model.add(Dense(1000, activation='softmax'))

    return model


def get_batches(directory, target_size=target_size, batch_size=batch_size, shuffle=False):
    datagen = ImageDataGenerator()
    return datagen.flow_from_directory(directory=directory,
                                          target_size=target_size,
                                          batch_size=batch_size,
                                          class_mode='categorical',
                                          shuffle=shuffle)
  

def get_weights(path, download=False):
    urllib.urlretrieve ("https://github.com/fchollet/deep-learning-models/releases/download/v0.1/vgg16_weights_tf_dim_ordering_tf_kernels.h5", model_path)

  
batches = get_batches(path+'train', shuffle=True)
valid_batches = get_batches(path+'valid', batch_size=batch_size*2, shuffle=False)


initial_model = create_vgg16()
initial_model.load_weights(model_path)



x = Dense(batches.num_class, activation='softmax')(initial_model.layers[-2].output)
model = Model(initial_model.input, x)

for layer in initial_model.layers: layer.trainable=False

opt = Adam(lr=0.001)
model.compile(optimizer=opt,
              loss='categorical_crossentropy', 
              metrics=['accuracy'])

model.fit_generator(batches, steps_per_epoch=batches.samples//batch_size, nb_epoch=1,
                validation_data=valid_batches, validation_steps=valid_batches.samples//batch_size)


for layer in model.layers[:10]:
   layer.trainable = False
for layer in model.layers[10:]:
   layer.trainable = True

opt = SGD(lr=10e-5)
model.compile(optimizer=opt,
              loss='categorical_crossentropy', 
              metrics=['accuracy'])

model.fit_generator(batches, steps_per_epoch=batches.samples//batch_size, nb_epoch=2,
                    validation_data=valid_batches, validation_steps=valid_batches.samples//batch_size)

hi,
If i run the code i am getting the following error:

2018-05-10 14:35:48.197082: I tensorflow/core/platform/cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX
Traceback (most recent call last):
File "ex3.py", line 89, in
x = Dense(batches.num_class, activation='softmax')(initial_model.layers[-2].output)
AttributeError: 'DirectoryIterator' object has no attribute 'num_class'

can u plz help me to solve this error

Thank you..

@jaggi did you fixed your bug?