web-dev-qa-db-fra.com

Comment utiliser le modèle de keras.applications pour transférer l'apprentissage?

Je veux obtenir un modèle VGG16 pré-formé dans Keras, supprimer sa couche de sortie, puis mettre une nouvelle couche de sortie avec le nombre de classes adaptées à mon problème, puis l'adapter à de nouvelles données. Pour cette raison, j'essaie d'utiliser le modèle ici: https://keras.io/applications/#vgg16 , mais comme il n'est pas séquentiel, je ne peux pas simplement model.pop() . Le saut à partir des calques et son ajout ne fonctionnent pas non plus, car dans les prédictions, il attend toujours l'ancienne forme. Comment ferais-je ça? Existe-t-il un moyen de convertir ce type de modèle en Sequential?

pythondeep-learningkeras
16
Hristo Vrigazov

Vous pouvez utiliser pop() sur model.layers puis utilisez model.layers[-1].output pour créer de nouveaux calques.

Exemple:

from keras.models import Model
from keras.layers import Dense,Flatten
from keras.applications import vgg16
from keras import backend as K

model = vgg16.VGG16(weights='imagenet', include_top=True)

model.input

model.summary(line_length=150)

model.layers.pop()
model.layers.pop()

model.summary(line_length=150)

new_layer = Dense(10, activation='softmax', name='my_dense')

inp = model.input
out = new_layer(model.layers[-1].output)

model2 = Model(inp, out)
model2.summary(line_length=150)

Vous pouvez également utiliser include_top=False option de ces modèles. Dans ce cas, si vous devez utiliser l'aplatissement du calque, vous devez passer le input_shape aussi.

model3 = vgg16.VGG16(weights='imagenet', include_top=False, input_shape=(224, 224, 3))
model3.summary(line_length=150)

flatten = Flatten()
new_layer2 = Dense(10, activation='softmax', name='my_dense_2')

inp2 = model3.input
out2 = new_layer2(flatten(model3.output))

model4 = Model(inp2, out2)
model4.summary(line_length=150)
46
indraforyou

Nous pouvons transformer le modèle VGG en séquentiel comme:

# Create VGG model
vgg_model = keras.applications.vgg16.VGG16(weights='imagenet')


# Created model is of type Model
type(vgg_model)
>> keras.engine.training.Model


# Convert it to Sequential
model = Sequential()
for layer in vgg_model.layers:
   model.add(layer)


# Now, check the model type, its Sequential! 
type(model)
>> keras.models.Sequential


# Verify the model details
model.summary()
>> 
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
input_15 (InputLayer)        (None, 224, 224, 3)       0         
_________________________________________________________________
block1_conv1 (Conv2D)        (None, 224, 224, 64)      1792      
_________________________________________________________________
block1_conv2 (Conv2D)        (None, 224, 224, 64)      36928     
_________________________________________________________________
block1_pool (MaxPooling2D)   (None, 112, 112, 64)      0         
_________________________________________________________________
block2_conv1 (Conv2D)        (None, 112, 112, 128)     73856     
_________________________________________________________________
block2_conv2 (Conv2D)        (None, 112, 112, 128)     147584    
_________________________________________________________________
block2_pool (MaxPooling2D)   (None, 56, 56, 128)       0         
_________________________________________________________________
block3_conv1 (Conv2D)        (None, 56, 56, 256)       295168    
_________________________________________________________________
block3_conv2 (Conv2D)        (None, 56, 56, 256)       590080    
_________________________________________________________________
block3_conv3 (Conv2D)        (None, 56, 56, 256)       590080    
_________________________________________________________________
block3_pool (MaxPooling2D)   (None, 28, 28, 256)       0         
_________________________________________________________________
block4_conv1 (Conv2D)        (None, 28, 28, 512)       1180160   
_________________________________________________________________
block4_conv2 (Conv2D)        (None, 28, 28, 512)       2359808   
_________________________________________________________________
block4_conv3 (Conv2D)        (None, 28, 28, 512)       2359808   
_________________________________________________________________
block4_pool (MaxPooling2D)   (None, 14, 14, 512)       0         
_________________________________________________________________
block5_conv1 (Conv2D)        (None, 14, 14, 512)       2359808   
_________________________________________________________________
block5_conv2 (Conv2D)        (None, 14, 14, 512)       2359808   
_________________________________________________________________
block5_conv3 (Conv2D)        (None, 14, 14, 512)       2359808   
_________________________________________________________________
block5_pool (MaxPooling2D)   (None, 7, 7, 512)         0         
_________________________________________________________________
flatten (Flatten)            (None, 25088)             0         
_________________________________________________________________
fc1 (Dense)                  (None, 4096)              102764544 
_________________________________________________________________
fc2 (Dense)                  (None, 4096)              16781312  
_________________________________________________________________
predictions (Dense)          (None, 1000)              4097000   
=================================================================
Total params: 138,357,544
Trainable params: 138,357,544
Non-trainable params: 0
_________________________________________________________________


# Now, that its sequential, we can perform usual operations.
model.layers.pop()


# Freeze the layers 
for layer in model.layers:
layer.trainable = False


# Add 'softmax' instead of earlier 'prediction' layer.
model.add(Dense(2, activation='softmax'))


# Check the summary, and yes new layer has been added. 
model.summary()

    Layer (type)                 Output Shape              Param #   
=================================================================
input_15 (InputLayer)        (None, 224, 224, 3)       0         
_________________________________________________________________
block1_conv1 (Conv2D)        (None, 224, 224, 64)      1792      
_________________________________________________________________
block1_conv2 (Conv2D)        (None, 224, 224, 64)      36928     
_________________________________________________________________
block1_pool (MaxPooling2D)   (None, 112, 112, 64)      0         
_________________________________________________________________
block2_conv1 (Conv2D)        (None, 112, 112, 128)     73856     
_________________________________________________________________
block2_conv2 (Conv2D)        (None, 112, 112, 128)     147584    
_________________________________________________________________
block2_pool (MaxPooling2D)   (None, 56, 56, 128)       0         
_________________________________________________________________
block3_conv1 (Conv2D)        (None, 56, 56, 256)       295168    
_________________________________________________________________
block3_conv2 (Conv2D)        (None, 56, 56, 256)       590080    
_________________________________________________________________
block3_conv3 (Conv2D)        (None, 56, 56, 256)       590080    
_________________________________________________________________
block3_pool (MaxPooling2D)   (None, 28, 28, 256)       0         
_________________________________________________________________
block4_conv1 (Conv2D)        (None, 28, 28, 512)       1180160   
_________________________________________________________________
block4_conv2 (Conv2D)        (None, 28, 28, 512)       2359808   
_________________________________________________________________
block4_conv3 (Conv2D)        (None, 28, 28, 512)       2359808   
_________________________________________________________________
block4_pool (MaxPooling2D)   (None, 14, 14, 512)       0         
_________________________________________________________________
block5_conv1 (Conv2D)        (None, 14, 14, 512)       2359808   
_________________________________________________________________
block5_conv2 (Conv2D)        (None, 14, 14, 512)       2359808   
_________________________________________________________________
block5_conv3 (Conv2D)        (None, 14, 14, 512)       2359808   
_________________________________________________________________
block5_pool (MaxPooling2D)   (None, 7, 7, 512)         0         
_________________________________________________________________
flatten (Flatten)            (None, 25088)             0         
_________________________________________________________________
fc1 (Dense)                  (None, 4096)              102764544 
_________________________________________________________________
fc2 (Dense)                  (None, 4096)              16781312  
_________________________________________________________________
dense_4 (Dense)              (None, 2)                 2002      
=================================================================
Total params: 134,262,546
Trainable params: 2,002
Non-trainable params: 134,260,544
_________________________________________________________________
3
Milind Deore