def layer_op(self, images, is_training=True, layer_id=-1, **unused_kwargs):
"""
:param images: tensor, concatenation of multiple input modalities
:param is_training: boolean, True if network is in training mode
:param layer_id: not in use
:param unused_kwargs:
:return: predicted tensor
"""
n_modality = images.shape.as_list()[-1]
rank = images.shape.ndims
assert n_modality > 1
roots = tf.split(images, n_modality, axis=rank - 1)
for (idx, root) in enumerate(roots):
conv_layer = ConvolutionalLayer(
n_output_chns=self.n_features,
kernel_size=3,
w_initializer=self.initializers['w'],
w_regularizer=self.regularizers['w'],
acti_func=self.acti_func,
name='conv_{}'.format(idx))
roots[idx] = conv_layer(root, is_training)
roots = tf.stack(roots, axis=-1)
back_end = ScaleBlock('AVERAGE', n_layers=1)
output_tensor = back_end(roots, is_training)
front_end = HighRes3DNet(self.num_classes)
output_tensor = front_end(output_tensor, is_training)
return output_tensor
def test_3d_shape(self):
input_shape = (2, 32, 32, 32, 1)
x = tf.ones(input_shape)
highres_layer = HighRes3DNet(num_classes=5)
out = highres_layer(x, is_training=True)
with self.test_session() as sess:
sess.run(tf.global_variables_initializer())
out = sess.run(out)
self.assertAllClose((2, 32, 32, 32, 5), out.shape)
def test_2d_reg_shape(self):
input_shape = (2, 32, 32, 1)
x = tf.ones(input_shape)
highres_layer = HighRes3DNet(
num_classes=5,
w_regularizer=regularizers.l2_regularizer(0.5),
b_regularizer=regularizers.l2_regularizer(0.5))
out = highres_layer(x, is_training=True)
# print(tf.get_collection(tf.GraphKeys.REGULARIZATION_LOSSES))
with self.test_session() as sess:
sess.run(tf.global_variables_initializer())
out = sess.run(out)
self.assertAllClose((2, 32, 32, 5), out.shape)
def shape_test(self, input_shape, expected_shape):
x = tf.ones(input_shape)
highres_layer = HighRes3DNet(num_classes=5)
highres_layer_small = HighRes3DNetSmall(num_classes=5)
highres_layer_large = HighRes3DNetLarge(num_classes=5)
out = highres_layer(x, is_training=True)
out_small = highres_layer_small(x, is_training=True)
out_large = highres_layer_large(x, is_training=True)
with self.cached_session() as sess:
sess.run(tf.global_variables_initializer())
out, out_large, out_small = sess.run([out, out_large, out_small])
self.assertAllClose(expected_shape, out.shape)
self.assertAllClose(expected_shape, out_large.shape)
self.assertAllClose(expected_shape, out_small.shape)
def shape_test_reg(self, input_shape, expected_shape):
x = tf.ones(input_shape)
layer_param = {
'num_classes': 5,
'w_regularizer': regularizers.l2_regularizer(0.5),
'b_regularizer': regularizers.l2_regularizer(0.5)}
highres_layer = HighRes3DNet(**layer_param)
highres_layer_small = HighRes3DNetSmall(**layer_param)
highres_layer_large = HighRes3DNetLarge(**layer_param)
out = highres_layer(x, is_training=True)
out_small = highres_layer_small(x, is_training=True)
out_large = highres_layer_large(x, is_training=True)
with self.cached_session() as sess:
sess.run(tf.global_variables_initializer())
out, out_large, out_small = sess.run([out, out_large, out_small])
self.assertAllClose(expected_shape, out.shape)
self.assertAllClose(expected_shape, out_large.shape)
self.assertAllClose(expected_shape, out_small.shape)
def layer_op(self, images, is_training=True, layer_id=-1, **unused_kwargs):
n_modality = images.shape.as_list()[-1]
rank = images.shape.ndims
assert n_modality > 1
roots = tf.split(images, n_modality, axis=rank - 1)
for (idx, root) in enumerate(roots):
conv_layer = ConvolutionalLayer(
n_output_chns=self.n_features,
kernel_size=3,
w_initializer=self.initializers['w'],
w_regularizer=self.regularizers['w'],
acti_func=self.acti_func,
name='conv_{}'.format(idx))
roots[idx] = conv_layer(root, is_training)
roots = tf.stack(roots, axis=-1)
back_end = ScaleBlock('AVERAGE', n_layers=1)
output_tensor = back_end(roots, is_training)
front_end = HighRes3DNet(self.num_classes)
output_tensor = front_end(output_tensor, is_training)
return output_tensor