We shall now use the class StyleContentModel defined in the previous slide.
When called on an image, this model returns the gram matrix (style) of the style_layers and content of the content_layers:
Note:
tf.constant() creates a constant tensor from a tensor-like object.
The output returned by the result items are of type tensorflow.python.framework.ops.EagerTensor. So we shall convert that into NumPy array and find the mathematical statistics - like minimum value, mean value, etc - of each output.
Instantiate the class StyleContentModel and pass the style_layers and content_layers.
extractor = << your code comes here >>(style_layers, content_layers)
Now pass the tf.constant(content_image) to the extractor and get the results - which holds the content and style representations of the content_image.
results = << your code comes here >>(tf.constant(content_image))
Let us see the statistics of the gram-matrices returned for the content image.
for name, output in sorted(results['style'].items()):
print(" ", name)
print(" shape: ", output.numpy().shape)
print(" min: ", output.numpy().min())
print(" max: ", output.numpy().max())
print(" mean: ", output.numpy().mean())
print()
Let us see the statistics of the content matrices returned for the content image.
for name, output in sorted(results['content'].items()):
print(" ", name)
print(" shape: ", output.numpy().shape)
print(" min: ", output.numpy().min())
print(" max: ", output.numpy().max())
print(" mean: ", output.numpy().mean())
Note - Having trouble with the assessment engine? Follow the steps listed here
Loading comments...