In [1]:

    

import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline
from keras.utils import np_utils
from keras.preprocessing.image import ImageDataGenerator
from keras.models import Sequential
from keras.layers import Dense, Dropout, Activation, Flatten
from keras.layers import Convolution2D, MaxPooling2D, AveragePooling2D
from keras.optimizers import Adam
import glob
from PIL import Image

import keras
from keras.preprocessing.image import ImageDataGenerator
from keras.models import Sequential
from keras.layers import Dense, Dropout, Activation, Flatten
from keras.layers import Conv2D, MaxPooling2D
from keras.utils import np_utils
from keras.layers.core import Flatten, Dense, Dropout, Lambda


    

    




Using TensorFlow backend.

In [ ]:

    




    

In [2]:

    

def plots(ims, figsize=(12,6), rows=1, interp=False, titles=None):
    if type(ims[0]) is np.ndarray:
        ims = np.array(ims).astype(np.uint8)
        if (ims.shape[-1] != 3):
            ims = ims.transpose((0,2,3,1))
    f = plt.figure(figsize=figsize)
    for i in range(len(ims)):
        sp = f.add_subplot(rows, len(ims)//rows, i+1)
        sp.axis('Off')
        if titles is not None:
            sp.set_title(titles[i], fontsize=16)
        plt.imshow(ims[i], interpolation=None if interp else 'none')


    

In [3]:

    

from keras.preprocessing import image

BATCH_SIZE = 64
PATH="data/"

def get_fit_sample():
    gen = image.ImageDataGenerator()
    sample_batches = gen.flow_from_directory(PATH+'valid', target_size=(224,224), 
            class_mode='categorical', shuffle=False, batch_size=300)
    imgs, labels = next(sample_batches)
    return imgs

gen = image.ImageDataGenerator(featurewise_std_normalization=True)
gen.fit(get_fit_sample())
val_batches = gen.flow_from_directory(PATH+'valid', target_size=(224,224), 
            class_mode='categorical', shuffle=True, batch_size=BATCH_SIZE)

gen = image.ImageDataGenerator(featurewise_std_normalization=True)
gen.fit(get_fit_sample())
batches = gen.flow_from_directory(PATH+'train', target_size=(224,224), 
            class_mode='categorical', shuffle=True, batch_size=BATCH_SIZE)

#imgs,labels = next(batches)
#plots(imgs[:2])


    

    




Found 2000 images belonging to 2 classes.
Found 2000 images belonging to 2 classes.
Found 2000 images belonging to 2 classes.
Found 16997 images belonging to 2 classes.

In [ ]:

    




    

In [4]:

    

CLASSES = 2
INPUT_SHAPE = (224,224,3)
model = Sequential()
    
# Block 1
model.add(Conv2D(64, (3, 3), activation='relu', padding='same', name='block1_conv1', input_shape=INPUT_SHAPE))
model.add(Conv2D(64, (3, 3), activation='relu', padding='same', name='block1_conv2'))
model.add(MaxPooling2D((2, 2), strides=(2, 2), name='block1_pool'))

# Block 2
model.add(Conv2D(128, (3, 3), activation='relu', padding='same', name='block2_conv1'))
model.add(Conv2D(128, (3, 3), activation='relu', padding='same', name='block2_conv2'))
model.add(MaxPooling2D((2, 2), strides=(2, 2), name='block2_pool'))

# Block 3
model.add(Conv2D(256, (3, 3), activation='relu', padding='same', name='block3_conv1'))
model.add(Conv2D(256, (3, 3), activation='relu', padding='same', name='block3_conv2'))
model.add(MaxPooling2D((2, 2), strides=(2, 2), name='block3_pool'))

# Block 4
model.add(Conv2D(512, (3, 3), activation='relu', padding='same', name='block4_conv1'))
model.add(Conv2D(512, (3, 3), activation='relu', padding='same', name='block4_conv2'))
model.add(MaxPooling2D((2, 2), strides=(2, 2), name='block4_pool'))

# Block 5
model.add(Conv2D(512, (3, 3), activation='relu', padding='same', name='block5_conv1'))
model.add(Conv2D(512, (3, 3), activation='relu', padding='same', name='block5_conv2'))
model.add(MaxPooling2D((2, 2), strides=(2, 2), name='block5_pool'))

# Classification block
model.add(Flatten(name='flatten'))
model.add(Dense(4096, activation='relu', name='fc1'))
model.add(Dropout(0.5))
model.add(Dense(4096, activation='relu', name='fc2'))
model.add(Dropout(0.5))
model.add(Dense(CLASSES, activation='softmax', name='predictions'))

from keras.optimizers import SGD
sgd = SGD(lr=0.01, decay=0.0005, momentum=0.9, nesterov=False)


model.compile(optimizer=sgd, loss='mean_squared_error', metrics=['accuracy'])


    

In [5]:

    

%%time
hist = model.fit_generator(batches,  steps_per_epoch=100, epochs=10, validation_data=val_batches, validation_steps=10)
 
model.save('ConvNet-B-vgg13.h5')

# http://qiita.com/TypeNULL/items/4e4d7de11ab4361d6085
loss = hist.history['loss']
val_loss = hist.history['val_loss']
nb_epoch = len(loss)
plt.plot(range(nb_epoch), loss, marker='.', label='loss')
plt.plot(range(nb_epoch), val_loss, marker='.', label='val_loss')
plt.legend(loc='best', fontsize=10)
plt.grid()
plt.xlabel('epoch')
plt.ylabel('loss')
plt.show()


    

    




Epoch 1/10
100/100 [==============================] - 93s - loss: 0.2492 - acc: 0.5233 - val_loss: 0.2474 - val_acc: 0.5781
Epoch 2/10
100/100 [==============================] - 89s - loss: 0.2453 - acc: 0.5631 - val_loss: 0.2450 - val_acc: 0.5375
Epoch 3/10
100/100 [==============================] - 92s - loss: 0.2414 - acc: 0.5705 - val_loss: 0.2352 - val_acc: 0.6031
Epoch 4/10
100/100 [==============================] - 90s - loss: 0.2330 - acc: 0.5923 - val_loss: 0.2271 - val_acc: 0.6453
Epoch 5/10
100/100 [==============================] - 90s - loss: 0.2255 - acc: 0.6236 - val_loss: 0.2257 - val_acc: 0.6188
Epoch 6/10
100/100 [==============================] - 89s - loss: 0.2193 - acc: 0.6306 - val_loss: 0.2100 - val_acc: 0.6500
Epoch 7/10
100/100 [==============================] - 89s - loss: 0.2078 - acc: 0.6677 - val_loss: 0.2085 - val_acc: 0.6622
Epoch 8/10
100/100 [==============================] - 89s - loss: 0.2043 - acc: 0.6762 - val_loss: 0.2035 - val_acc: 0.6641
Epoch 9/10
100/100 [==============================] - 89s - loss: 0.1913 - acc: 0.7091 - val_loss: 0.1825 - val_acc: 0.7359
Epoch 10/10
100/100 [==============================] - 89s - loss: 0.1907 - acc: 0.7013 - val_loss: 0.1745 - val_acc: 0.7250






    













    




CPU times: user 10min 54s, sys: 1min 44s, total: 12min 38s
Wall time: 15min 5s

In [ ]: