def cnn(x):
with tf.name_scope('cnn') as scope:
with tf.name_scope('cnn_conv1') as inner_scope:
wcnn1 = tu.weight([11, 11, 3, 96], name='wcnn1')
bcnn1 = tu.bias(0.0, [96], name='bcnn1')
conv1 = tf.add(tu.conv2d(x, wcnn1, stride=(4, 4), padding='SAME'),
bcnn1)
conv1 = tu.relu(conv1)
norm1 = tu.lrn(conv1,
depth_radius=2,
bias=1.0,
alpha=2e-05,
beta=0.75)
pool1 = tu.max_pool2d(norm1,
kernel=[1, 3, 3, 1],
stride=[1, 2, 2, 1],
padding='VALID')
with tf.name_scope('cnn_conv2') as inner_scope:
wcnn2 = tu.weight([5, 5, 96, 256], name='wcnn2')
bcnn2 = tu.bias(1.0, [256], name='bcnn2')
conv2 = tf.add(
tu.conv2d(pool1, wcnn2, stride=(1, 1), padding='SAME'), bcnn2)
conv2 = tu.relu(conv2)
norm2 = tu.lrn(conv2,
depth_radius=2,
bias=1.0,
alpha=2e-05,
beta=0.75)
pool2 = tu.max_pool2d(norm2,
kernel=[1, 3, 3, 1],
stride=[1, 2, 2, 1],
padding='VALID')
with tf.name_scope('cnn_conv3') as inner_scope:
wcnn3 = tu.weight([3, 3, 256, 384], name='wcnn3')
bcnn3 = tu.bias(0.0, [384], name='bcnn3')
conv3 = tf.add(
tu.conv2d(pool2, wcnn3, stride=(1, 1), padding='SAME'), bcnn3)
conv3 = tu.relu(conv3)
with tf.name_scope('cnn_conv4') as inner_scope:
wcnn4 = tu.weight([3, 3, 384, 384], name='wcnn4')
bcnn4 = tu.bias(1.0, [384], name='bcnn4')
conv4 = tf.add(
tu.conv2d(conv3, wcnn4, stride=(1, 1), padding='SAME'), bcnn4)
conv4 = tu.relu(conv4)
with tf.name_scope('cnn_conv5') as inner_scope:
wcnn5 = tu.weight([3, 3, 384, 256], name='wcnn5')
bcnn5 = tu.bias(1.0, [256], name='bcnn5')
conv5 = tf.add(
tu.conv2d(conv4, wcnn5, stride=(1, 1), padding='SAME'), bcnn5)
conv5 = tu.relu(conv5)
pool5 = tu.max_pool2d(conv5,
kernel=[1, 3, 3, 1],
stride=[1, 2, 2, 1],
padding='VALID')
return pool5
def cnn(x):
"""
AlexNet convolutional layers definition
Args:
x: tensor of shape [batch_size, width, height, channels]
Returns:
pool5: tensor with all convolutions, pooling and lrn operations applied
"""
with tf.name_scope('alexnet_cnn') as scope:
with tf.name_scope('alexnet_cnn_conv1') as inner_scope:
wcnn1 = tu.weight([11, 11, 3, 96], name='wcnn1')
bcnn1 = tu.bias(0.0, [96], name='bcnn1')
conv1 = tf.add(tu.conv2d(x, wcnn1, stride=(4, 4), padding='SAME'), bcnn1)
# conv1 = tu.batch_norm(conv1)
conv1 = tu.relu(conv1)
norm1 = tu.lrn(conv1, depth_radius=2, bias=1.0, alpha=2e-05, beta=0.75)
pool1 = tu.max_pool2d(norm1, kernel=[1, 3, 3, 1], stride=[1, 2, 2, 1], padding='VALID')
with tf.name_scope('alexnet_cnn_conv2') as inner_scope:
wcnn2 = tu.weight([5, 5, 96, 256], name='wcnn2')
bcnn2 = tu.bias(1.0, [256], name='bcnn2')
conv2 = tf.add(tu.conv2d(pool1, wcnn2, stride=(1, 1), padding='SAME'), bcnn2)
# conv2 = tu.batch_norm(conv2)
conv2 = tu.relu(conv2)
norm2 = tu.lrn(conv2, depth_radius=2, bias=1.0, alpha=2e-05, beta=0.75)
pool2 = tu.max_pool2d(norm2, kernel=[1, 3, 3, 1], stride=[1, 2, 2, 1], padding='VALID')
with tf.name_scope('alexnet_cnn_conv3') as inner_scope:
wcnn3 = tu.weight([3, 3, 256, 384], name='wcnn3')
bcnn3 = tu.bias(0.0, [384], name='bcnn3')
conv3 = tf.add(tu.conv2d(pool2, wcnn3, stride=(1, 1), padding='SAME'), bcnn3)
# conv3 = tu.batch_norm(conv3)
conv3 = tu.relu(conv3)
with tf.name_scope('alexnet_cnn_conv4') as inner_scope:
wcnn4 = tu.weight([3, 3, 384, 384], name='wcnn4')
bcnn4 = tu.bias(1.0, [384], name='bcnn4')
conv4 = tf.add(tu.conv2d(conv3, wcnn4, stride=(1, 1), padding='SAME'), bcnn4)
# conv4 = tu.batch_norm(conv4)
conv4 = tu.relu(conv4)
with tf.name_scope('alexnet_cnn_conv5') as inner_scope:
wcnn5 = tu.weight([3, 3, 384, 256], name='wcnn5')
bcnn5 = tu.bias(1.0, [256], name='bcnn5')
conv5 = tf.add(tu.conv2d(conv4, wcnn5, stride=(1, 1), padding='SAME'), bcnn5)
# conv5 = tu.batch_norm(conv5)
conv5 = tu.relu(conv5)
pool5 = tu.max_pool2d(conv5, kernel=[1, 3, 3, 1], stride=[1, 2, 2, 1], padding='VALID')
return pool5
def cnn(x):
"""
AlexNet convolutional layers definition
Args:
x: tensor of shape [batch_size, width, height, channels]
Returns:
pool5: tensor with all convolutions, pooling and lrn operations applied
"""
with tf.name_scope('alexnet_cnn') as scope:
with tf.name_scope('alexnet_cnn_conv1') as inner_scope:
wcnn1 = tu.weight([11, 11, 3, 96], name='wcnn1')
# bcnn1 = tu.bias(0.0, [96], name='bcnn1')
# wcnn1_t = fw(wcnn1)
# x_t =fa(cabs(x))
conv1 = tu.conv2d(x, wcnn1, stride=(4, 4), padding='SAME')
#conv1 = tu.batch_norm(conv1)
conv1 = tf.nn.relu(conv1)
norm1 = tu.lrn(conv1,
depth_radius=2,
bias=1.0,
alpha=2e-05,
beta=0.75)
pool1 = tu.max_pool2d(norm1,
kernel=[1, 3, 3, 1],
stride=[1, 2, 2, 1],
padding='VALID')
with tf.name_scope('alexnet_cnn_conv2') as inner_scope:
wcnn2 = tu.weight([5, 5, 96, 256], name='wcnn2')
# bcnn2 = tu.bias(1.0, [256], name='bcnn2')
pool1_t = fa(cabs(pool1))
wcnn2_t = fw(wcnn2)
conv2 = tu.conv2d(pool1_t, wcnn2_t, stride=(1, 1), padding='SAME')
#conv2 = tu.batch_norm(conv2)
conv2 = tf.nn.relu(conv2)
norm2 = tu.lrn(conv2,
depth_radius=2,
bias=1.0,
alpha=2e-05,
beta=0.75)
pool2 = tu.max_pool2d(norm2,
kernel=[1, 3, 3, 1],
stride=[1, 2, 2, 1],
padding='VALID')
with tf.name_scope('alexnet_cnn_conv3') as inner_scope:
wcnn3 = tu.weight([3, 3, 256, 384], name='wcnn3')
# bcnn3 = tu.bias(0.0, [384], name='bcnn3')
pool2_t = fa(cabs(pool2))
wcnn3_t = fw(wcnn3)
conv3 = tu.conv2d(pool2_t, wcnn3_t, stride=(1, 1), padding='SAME')
#conv3 = tu.batch_norm(conv3)
conv3 = tf.nn.relu(conv3)
with tf.name_scope('alexnet_cnn_conv4') as inner_scope:
wcnn4 = tu.weight([3, 3, 384, 384], name='wcnn4')
# bcnn4 = tu.bias(1.0, [384], name='bcnn4')
conv3_t = fa(cabs(conv3))
wcnn4_t = fw(wcnn4)
conv4 = tu.conv2d(conv3_t, wcnn4_t, stride=(1, 1), padding='SAME')
#conv4 = tu.batch_norm(conv4)
conv4 = tf.nn.relu(conv4)
with tf.name_scope('alexnet_cnn_conv5') as inner_scope:
wcnn5 = tu.weight([3, 3, 384, 256], name='wcnn5')
# bcnn5 = tu.bias(1.0, [256], name='bcnn5')
conv4_t = fa(cabs(conv4))
wcnn5_t = fw(wcnn5)
conv5 = tu.conv2d(conv4_t, wcnn5_t, stride=(1, 1), padding='SAME')
#conv5 = tu.batch_norm(conv5)
conv5 = tf.nn.relu(conv5)
pool5 = tu.max_pool2d(conv5,
kernel=[1, 3, 3, 1],
stride=[1, 2, 2, 1],
padding='VALID')
return pool5