파이썬 RuntimeError Variable += value not supported. Use variable.assign_add(value) to modify the variable value and variable = variable + value to get a new Tensor object.

케라스 창시자에게 배우는 딥러닝 실습 예제 8.3-neural-style-transfer 을 돌려 보던 중 런타임에러가 떴다.

 

 

# Dict mapping layer names to activation tensors
outputs_dict = dict([(layer.name, layer.output) for layer in model.layers])
# Name of layer used for content loss
content_layer = 'block5_conv2'
# Name of layers used for style loss
style_layers = ['block1_conv1',
                'block2_conv1',
                'block3_conv1',
                'block4_conv1',
                'block5_conv1']
# Weights in the weighted average of the loss components
total_variation_weight = 1e-4
style_weight = 1.
content_weight = 0.025

# Define the loss by adding all components to a `loss` variable
loss = K.variable(0.)
layer_features = outputs_dict[content_layer]
target_image_features = layer_features[0, :, :, :]
combination_features = layer_features[2, :, :, :]
loss += content_weight * content_loss(target_image_features,
                                      combination_features)
for layer_name in style_layers:
    layer_features = outputs_dict[layer_name]
    style_reference_features = layer_features[1, :, :, :]
    combination_features = layer_features[2, :, :, :]
    sl = style_loss(style_reference_features, combination_features)
    loss += (style_weight / len(style_layers)) * sl
loss += total_variation_weight * total_variation_loss(combination_image)

 

RuntimeError: Variable += value not supported. Use variable.assign_add(value) to modify the variable value and variable = variable + value to get a new Tensor object.

 

loss를 다음과 같이 수정해주었더니 잘 돌아간다.

 

# Dict mapping layer names to activation tensors
outputs_dict = dict([(layer.name, layer.output) for layer in model.layers])
# Name of layer used for content loss
content_layer = 'block5_conv2'
# Name of layers used for style loss
style_layers = ['block1_conv1',
                'block2_conv1',
                'block3_conv1',
                'block4_conv1',
                'block5_conv1']
# Weights in the weighted average of the loss components
total_variation_weight = 1e-4
style_weight = 1.
content_weight = 0.025

# Define the loss by adding all components to a `loss` variable
loss = K.variable(0.)
layer_features = outputs_dict[content_layer]
target_image_features = layer_features[0, :, :, :]
combination_features = layer_features[2, :, :, :]
loss = loss + content_weight * content_loss(target_image_features,
                                      combination_features)
for layer_name in style_layers:
    layer_features = outputs_dict[layer_name]
    style_reference_features = layer_features[1, :, :, :]
    combination_features = layer_features[2, :, :, :]
    sl = style_loss(style_reference_features, combination_features)
    loss += (style_weight / len(style_layers)) * sl
loss += total_variation_weight * total_variation_loss(combination_image)

 

 

 

tensorflow.keras가 아닌 그냥 keras로 사용했을때도 위와 같은 에러가 발생 하나보다. (tensorflow가 2.xx 로 업뎃 되면서 keras가 텐서플로 안으로 들어오게 됐다.) 하지만 난 처음부터 tensorflow.keras로 진행했음에도 위와 같은 에러가 발생해 loss를 수정한 것이다.

 

 

또한 ImportError: cannot import name 'imsave' 가 발생했는데 from scipy.misc import imsave 대신 from keras.preprocessing.image import save_img 를 써주면 해결된다.

 

 

from scipy.optimize import fmin_l_bfgs_b
from tensorflow.keras.preprocessing.image import save_img
import time

result_prefix = 'style_transfer_result'
iterations = 20

# Run scipy-based optimization (L-BFGS) over the pixels of the generated image
# so as to minimize the neural style loss.
# This is our initial state: the target image.
# Note that `scipy.optimize.fmin_l_bfgs_b` can only process flat vectors.
x = preprocess_image(target_image_path)
x = x.flatten()
for i in range(iterations):
    print('Start of iteration', i)
    start_time = time.time()
    x, min_val, info = fmin_l_bfgs_b(evaluator.loss, x,
                                     fprime=evaluator.grads, maxfun=20)
    print('Current loss value:', min_val)
    # Save current generated image
    img = x.copy().reshape((img_height, img_width, 3))
    img = deprocess_image(img)
    fname = result_prefix + '_at_iteration_%d.png' % i
    save_img(fname, img)
    
    end_time = time.time()
    print('Image saved as', fname)
    print('Iteration %d completed in %ds' % (i, end_time - start_time))

 

 

 

그럼 아래와 같은 combination image를 얻게 된다!!