def decoder(inputs):
# decoder
# 4 x 4 x 4 x 8 -> 8 x 8 x 8 x 16
# 8 x 8 x 8 x 16 -> 16 x 16 x 16 x 32
# 16 x 16 x 16 x 32 -> 32 x 32 x 32 x 1
#net = lays.fully_connected(inputs, 1)
#print("decoder input shape", inputs[1], tf.shape(inputs))
# net = lays.conv3d_transpose(net, 8, [4, 4, 4], stride=2, padding='SAME', trainable=True)
net = tf.layers.dense(inputs, units=512, activation=tf.nn.relu)
net = tf.reshape(net, [-1, 4, 4, 4, 8])
print("Here", net, tf.shape(net))
net = lays.conv3d_transpose(net,
16, [4, 4, 4],
stride=2,
padding='SAME',
trainable=True)
net = lays.batch_norm(net, decay=0.999)
net = lays.conv3d_transpose(net,
32, [4, 4, 4],
stride=2,
padding='SAME',
trainable=True)
net = lays.batch_norm(net, decay=0.999)
net = lays.conv3d_transpose(net,
1, [4, 4, 4],
stride=2,
padding='SAME',
activation_fn=tf.nn.sigmoid,
trainable=True)
return net
def decoder(inputs):
# decoder
# 2 x 2 x 2 x 8 -> 8 x 8 x 8 x 16
# 8 x 8 x 8 x 16 -> 16 x 16 x 16 x 32
# 16 x 16 x 16 x 32 -> 32 x 32 x 32 x 1
net = tf.layers.dense(inputs, 2 * 2 * 2 * 8, activation=tf.nn.relu)
#net = tf.layers.dense(net, 2 * 2 * 2 * 8, activation=tf.nn.relu)
net = tf.reshape(net, [-1, 2, 2, 2, 8])
net = lays.batch_norm(lays.conv3d_transpose(net,
16, [5, 5, 5],
stride=4,
padding='SAME',
trainable=True),
decay=0.9)
net = lays.batch_norm(lays.conv3d_transpose(net,
32, [5, 5, 5],
stride=2,
padding='SAME',
trainable=True),
decay=0.9)
net = lays.conv3d_transpose(net,
1, [5, 5, 5],
stride=2,
padding='SAME',
activation_fn=tf.nn.sigmoid,
trainable=True)
return net
def _3d_rdnn_level_60(self, inputs):
with tf.variable_scope('Volume30Up', reuse=self._reuse):
return layers.conv3d_transpose(inputs,
256,
3,
2,
scope='deconv',
**self._reg)
def autoencoder(inputs):
# encoder
# 32 x 32 x 32 x 1 -> 16 x 16 x 16 x 32
# 16 x 16 x 16 x 32 -> 8 x 8 x 8 x 16
# 8 x 8 x 8 x 16 -> 2 x 2 x 2 x 8
net = lays.conv3d(inputs, 32, [5, 5, 5], stride=2, padding='SAME')
net = lays.conv3d(net, 16, [5, 5, 5], stride=2, padding='SAME')
print(tf.shape(net))
net = lays.conv3d(net, 8, [5, 5, 5], stride=4, padding='SAME')
#net = lays.fully_connected(net,1)
latent_space = net
# decoder
# 2 x 2 x 2 x 8 -> 8 x 8 x 8 x 16
# 8 x 8 x 8 x 16 -> 16 x 16 x 16 x 32
# 16 x 16 x 16 x 32 -> 32 x 32 x 32 x 1
net = lays.conv3d_transpose(net, 16, [5, 5, 5], stride=4, padding='SAME')
net = lays.conv3d_transpose(net, 32, [5, 5, 5], stride=2, padding='SAME')
net = lays.conv3d_transpose(net, 1, [5, 5, 5], stride=2, padding='SAME', activation_fn=tf.nn.tanh)
return latent_space, net