def test_session(self):
vgg = VGG19(include_top=False, weights="imagenet")
vgg.trainable = False
session = tf.compat.v1.Session()
perceptual_vgg = smlosses.PerceptualVGG(vgg=vgg, eager=False, session=session)
from skimage import data
from supermariopy.tfutils import image
x = (
data.astronaut()
.astype(np.float32)
.reshape((1, 512, 512, 3))[:, :224, :224, :]
)
y = (
data.astronaut()
.astype(np.float32)
.reshape((1, 512, 512, 3))[:, :224, :224, :]
)
x_ph = tf.placeholder(shape=(1, 224, 224, 3), dtype=tf.float32)
y_ph = tf.placeholder(shape=(1, 224, 224, 3), dtype=tf.float32)
loss = perceptual_vgg.loss(x_ph, y_ph)
session.run(tf.initialize_all_variables())
loss_v = session.run(loss, {x_ph: x, y_ph: y})
assert all([np.allclose(l, np.array([0])) for l in loss_v])
def __init__(self,
session,
feature_layers=None,
feature_weights=None,
gram_weights=None):
K.set_session(session)
self.base_model = VGG19(include_top=False, weights='imagenet')
if feature_layers is None:
feature_layers = [
"input_1", "block1_conv2", "block2_conv2", "block3_conv2",
"block4_conv2", "block5_conv2"
]
self.layer_names = [l.val_set_name for l in self.base_model.layers]
for k in feature_layers:
if not k in self.layer_names:
raise KeyError("Invalid layer {}. Available layers: {}".format(
k, self.layer_names))
features = [
self.base_model.get_layer(k).output for k in feature_layers
]
self.model = Model(inputs=self.base_model.input, outputs=features)
if feature_weights is None:
feature_weights = len(feature_layers) * [1.0]
if gram_weights is None:
gram_weights = len(feature_layers) * [0.1]
self.feature_weights = feature_weights
self.gram_weights = gram_weights
assert len(self.feature_weights) == len(features)
self.use_gram = np.max(self.gram_weights) > 0.0
self.variables = self.base_model.weights
def test_eager(self):
tf.enable_eager_execution()
vgg = VGG19(include_top=False, weights="imagenet")
vgg.trainable = False
perceptual_vgg = smlosses.PerceptualVGG(vgg=vgg, eager=True)
x = data.astronaut().astype(np.float32).reshape((1, 512, 512, 3))
x = image.resize_bilinear(x, [224, 224])
y = data.astronaut().astype(np.float32).reshape((1, 512, 512, 3))
y = image.resize_bilinear(y, [224, 224])
losses = perceptual_vgg.loss(tf.convert_to_tensor(x), tf.convert_to_tensor(y))
assert all([np.allclose(loss, np.array([0])) for loss in losses])
y = image.resize_bilinear(y, [224, 224]) + tf.random.normal(y.shape)
losses = perceptual_vgg.loss(tf.convert_to_tensor(x), tf.convert_to_tensor(y))
assert all([not np.allclose(loss, np.array([0])) for loss in losses])
def __init__(
self,
session,
feature_layers=None,
feature_weights=None,
gram_weights=None,
default_gram=0.1,
original_scale=False,
eager=False,
):
if eager:
pass
else:
K.set_session(session)
self.base_model = VGG19(include_top=False, weights="imagenet")
if feature_layers is None:
feature_layers = [
"input_1",
"block1_conv2",
"block2_conv2",
"block3_conv2",
"block4_conv2",
"block5_conv2",
]
self.layer_names = [l.name for l in self.base_model.layers]
for k in feature_layers:
if not k in self.layer_names:
raise KeyError("Invalid layer {}. Available layers: {}".format(
k, self.layer_names))
self.feature_layers = feature_layers
features = [
self.base_model.get_layer(k).output for k in feature_layers
]
self.model = Model(inputs=self.base_model.input, outputs=features)
if feature_weights is None:
feature_weights = len(feature_layers) * [1.0]
if gram_weights is None:
gram_weights = len(feature_layers) * [default_gram]
elif isinstance(gram_weights, (int, float)):
gram_weights = len(feature_layers) * [gram_weights]
self.feature_weights = feature_weights
self.gram_weights = gram_weights
assert len(self.feature_weights) == len(features)
self.use_gram = np.max(self.gram_weights) > 0.0
self.original_scale = original_scale
self.variables = self.base_model.weights
def __init__(self):
# self.input = tf.placeholder(tf.float32, [None,224,224,3], name='input_img')
self.model = VGG19(include_top=False)