VGGNet in Keras

In [1]:

    

import numpy as np
np.random.seed(42)


    

In [2]:

    

import keras
from keras.models import Sequential
from keras.layers import Dense, Dropout, Flatten, Conv2D, MaxPooling2D
from keras.layers.normalization import BatchNormalization
from keras.callbacks import TensorBoard # for part 3.5 on TensorBoard


    

    




Using TensorFlow backend.

In [3]:

    

import tflearn.datasets.oxflower17 as oxflower17
X, Y = oxflower17.load_data(one_hot=True)


    

In [4]:

    

model = Sequential()

model.add(Conv2D(64, 3, activation='relu', input_shape=(224, 224, 3)))
model.add(Conv2D(64, 3, activation='relu'))
model.add(MaxPooling2D(2, 2))
model.add(BatchNormalization())

model.add(Conv2D(128, 3, activation='relu'))
model.add(Conv2D(128, 3, activation='relu'))
model.add(MaxPooling2D(2, 2))
model.add(BatchNormalization())

model.add(Conv2D(256, 3, activation='relu'))
model.add(Conv2D(256, 3, activation='relu'))
model.add(Conv2D(256, 3, activation='relu'))
model.add(MaxPooling2D(2, 2))
model.add(BatchNormalization())

model.add(Conv2D(512, 3, activation='relu'))
model.add(Conv2D(512, 3, activation='relu'))
model.add(Conv2D(512, 3, activation='relu'))
model.add(MaxPooling2D(2, 2))
model.add(BatchNormalization())

model.add(Conv2D(512, 3, activation='relu'))
model.add(Conv2D(512, 3, activation='relu'))
model.add(Conv2D(512, 3, activation='relu'))
model.add(MaxPooling2D(2, 2))
model.add(BatchNormalization())

model.add(Flatten())
model.add(Dense(4096, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(4096, activation='relu'))
model.add(Dropout(0.5))

model.add(Dense(17, activation='softmax'))


    

In [5]:

    

model.summary()


    

    




_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv2d_1 (Conv2D)            (None, 222, 222, 64)      1792      
_________________________________________________________________
conv2d_2 (Conv2D)            (None, 220, 220, 64)      36928     
_________________________________________________________________
max_pooling2d_1 (MaxPooling2 (None, 110, 110, 64)      0         
_________________________________________________________________
batch_normalization_1 (Batch (None, 110, 110, 64)      256       
_________________________________________________________________
conv2d_3 (Conv2D)            (None, 108, 108, 128)     73856     
_________________________________________________________________
conv2d_4 (Conv2D)            (None, 106, 106, 128)     147584    
_________________________________________________________________
max_pooling2d_2 (MaxPooling2 (None, 53, 53, 128)       0         
_________________________________________________________________
batch_normalization_2 (Batch (None, 53, 53, 128)       512       
_________________________________________________________________
conv2d_5 (Conv2D)            (None, 51, 51, 256)       295168    
_________________________________________________________________
conv2d_6 (Conv2D)            (None, 49, 49, 256)       590080    
_________________________________________________________________
conv2d_7 (Conv2D)            (None, 47, 47, 256)       590080    
_________________________________________________________________
max_pooling2d_3 (MaxPooling2 (None, 23, 23, 256)       0         
_________________________________________________________________
batch_normalization_3 (Batch (None, 23, 23, 256)       1024      
_________________________________________________________________
conv2d_8 (Conv2D)            (None, 21, 21, 512)       1180160   
_________________________________________________________________
conv2d_9 (Conv2D)            (None, 19, 19, 512)       2359808   
_________________________________________________________________
conv2d_10 (Conv2D)           (None, 17, 17, 512)       2359808   
_________________________________________________________________
max_pooling2d_4 (MaxPooling2 (None, 8, 8, 512)         0         
_________________________________________________________________
batch_normalization_4 (Batch (None, 8, 8, 512)         2048      
_________________________________________________________________
conv2d_11 (Conv2D)           (None, 6, 6, 512)         2359808   
_________________________________________________________________
conv2d_12 (Conv2D)           (None, 4, 4, 512)         2359808   
_________________________________________________________________
conv2d_13 (Conv2D)           (None, 2, 2, 512)         2359808   
_________________________________________________________________
max_pooling2d_5 (MaxPooling2 (None, 1, 1, 512)         0         
_________________________________________________________________
batch_normalization_5 (Batch (None, 1, 1, 512)         2048      
_________________________________________________________________
flatten_1 (Flatten)          (None, 512)               0         
_________________________________________________________________
dense_1 (Dense)              (None, 4096)              2101248   
_________________________________________________________________
dropout_1 (Dropout)          (None, 4096)              0         
_________________________________________________________________
dense_2 (Dense)              (None, 4096)              16781312  
_________________________________________________________________
dropout_2 (Dropout)          (None, 4096)              0         
_________________________________________________________________
dense_3 (Dense)              (None, 17)                69649     
=================================================================
Total params: 33,672,785
Trainable params: 33,669,841
Non-trainable params: 2,944
_________________________________________________________________

In [6]:

    

model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])


    

In [ ]:

    

tensorbrd = TensorBoard('logs/vggnet')


    

In [ ]:

    

model.fit(X, Y, batch_size=64, epochs=16, verbose=1, validation_split=0.1, shuffle=True,
          callbacks=[tensorbrd])


    

    




Train on 1224 samples, validate on 136 samples
Epoch 1/16
1224/1224 [==============================] - 784s - loss: 4.1356 - acc: 0.0735 - val_loss: 12.5035 - val_acc: 0.0368
Epoch 2/16
1224/1224 [==============================] - 774s - loss: 3.4962 - acc: 0.1062 - val_loss: 13.7122 - val_acc: 0.0735
Epoch 3/16
1224/1224 [==============================] - 773s - loss: 2.9282 - acc: 0.1373 - val_loss: 13.0526 - val_acc: 0.0441
Epoch 4/16
1224/1224 [==============================] - 774s - loss: 3.1878 - acc: 0.1381 - val_loss: 3.1900 - val_acc: 0.0956
Epoch 5/16
1224/1224 [==============================] - 773s - loss: 3.4662 - acc: 0.0752 - val_loss: 2.8582 - val_acc: 0.0588
Epoch 6/16
1224/1224 [==============================] - 770s - loss: 3.2509 - acc: 0.1119 - val_loss: 2.8061 - val_acc: 0.0441
Epoch 7/16
1224/1224 [==============================] - 770s - loss: 2.8882 - acc: 0.1193 - val_loss: 2.6081 - val_acc: 0.1029
Epoch 8/16
1224/1224 [==============================] - 772s - loss: 3.1455 - acc: 0.1144 - val_loss: 3.9969 - val_acc: 0.0588
Epoch 9/16
1224/1224 [==============================] - 773s - loss: 3.2672 - acc: 0.1185 - val_loss: 5.0395 - val_acc: 0.0515
Epoch 10/16
1224/1224 [==============================] - 771s - loss: 2.8959 - acc: 0.1324 - val_loss: 2.8850 - val_acc: 0.0662
Epoch 11/16
1224/1224 [==============================] - 771s - loss: 2.6121 - acc: 0.1765 - val_loss: 2.7125 - val_acc: 0.0588
Epoch 12/16
1224/1224 [==============================] - 771s - loss: 2.5183 - acc: 0.1675 - val_loss: 2.5720 - val_acc: 0.1176
Epoch 13/16
1224/1224 [==============================] - 771s - loss: 2.6153 - acc: 0.1822 - val_loss: 6.8114 - val_acc: 0.1103
Epoch 14/16
1224/1224 [==============================] - 770s - loss: 2.6700 - acc: 0.1748 - val_loss: 4.6305 - val_acc: 0.1324
Epoch 15/16
1088/1224 [=========================>....] - ETA: 82s - loss: 2.7326 - acc: 0.1967 

In [ ]: