def model(x, n_outputs):
"""
Function that builds the graph for the neural network.
:param x: Input data
:param n_outputs: Number of output neurons
:return: Output vector of network, keep probability, and weight matrices in the network
"""
# First convolutional layer
x_image = tf.reshape(x, [-1, 250, 250, 3])
w_conv1 = weight_variable([10, 10, 3, 6], name="w_conv1")
b_conv1 = bias_variable([6], name="b_conv1")
h_conv1 = tf.nn.relu(conv2d(x_image, w_conv1) + b_conv1)
# Second convolutional layer
w_conv2 = weight_variable([5, 5, 6, 16], name="w_conv2")
b_conv2 = bias_variable([16], name="b_conv2")
h_conv2 = tf.nn.relu(conv2d(h_conv1, w_conv2) + b_conv2)
# Fully connected layer
w_fc1 = weight_variable([237 * 237 * 16, 80], name="w_fc1")
b_fc1 = bias_variable([80], name="b_fc1")
h_conv2_flat = tf.reshape(h_conv2, [-1, 237 * 237 * 16])
h_fc1 = tf.nn.relu(tf.matmul(h_conv2_flat, w_fc1) + b_fc1)
# Dropout on the fully connected layer
dropout = tf.placeholder(tf.float32)
h_fc1_drop = tf.nn.dropout(h_fc1, rate=dropout)
# Output layer
w_fc2 = weight_variable([80, n_outputs], name="w_fc2")
b_fc2 = bias_variable([n_outputs], name="b_fc2")
y_conv = tf.matmul(h_fc1_drop, w_fc2) + b_fc2
# Returns the prediction and the dropout probability placeholder
return y_conv, dropout, w_conv1, w_conv2, w_fc1, w_fc2
def create_UGRNN_variable(self):
with tf.variable_scope(self._Unique_ID + "EncodingNN") as scope:
contextual_features = tf.get_variable(
"contextual_features", [
config.max_seq_len * config.max_seq_len * 4,
self.encoding_nn_output_size
],
dtype=tf.float32,
initializer=tf.constant_initializer(0),
trainable=False)
with tf.variable_scope('hidden1') as scope:
weights = nn_utils.weight_variable(
[
self.encoding_nn_input_size,
self.encoding_nn_hidden_size
],
initializer=self.initializer_fun)
biases = nn_utils.bias_variable([self.encoding_nn_hidden_size])
self.trainable_variables.append(weights)
self.trainable_variables.append(biases)
with tf.variable_scope('output') as scope:
weights = nn_utils.weight_variable(
[
self.encoding_nn_hidden_size,
self.encoding_nn_output_size
],
initializer=self.initializer_fun)
biases = nn_utils.bias_variable([self.encoding_nn_output_size])
self.trainable_variables.append(weights)
self.trainable_variables.append(biases)
with tf.variable_scope(self._Unique_ID + "OutputNN") as scope:
with tf.variable_scope('hidden1') as scope:
weights = nn_utils.weight_variable(
[self.output_nn_input_size, self.output_nn_hidden_size],
self.initializer_fun, 'weights_decay')
biases = nn_utils.bias_variable([self.output_nn_hidden_size])
self.trainable_variables.append(weights)
self.trainable_variables.append(biases)
with tf.variable_scope('output') as scope:
weights = nn_utils.weight_variable(
[self.output_nn_hidden_size, 1], self.initializer_fun,
'weights_decay')
self.trainable_variables.append(weights)
def build_graph_forLwF(self, sess):
with tf.variable_scope("AlexNet_ImageNet"):
train_x = np.zeros((1, 227, 227, 3)).astype(float32)
train_y = np.zeros((1, 1000))
xdim = train_x.shape[1:]
ydim = train_y.shape[1]
net_data = load("bvlc_alexnet.npy").item()
def conv(input,
kernel,
biases,
k_h,
k_w,
c_o,
s_h,
s_w,
padding="VALID",
group=1):
'''From https://github.com/ethereon/caffe-tensorflow
'''
c_i = input.get_shape()[-1]
assert c_i % group == 0
assert c_o % group == 0
convolve = lambda i, k: tf.nn.conv2d(
i, k, [1, s_h, s_w, 1], padding=padding)
if group == 1:
conv = convolve(input, kernel)
else:
input_groups = tf.split(3, group, input)
kernel_groups = tf.split(3, group, kernel)
output_groups = [
convolve(i, k)
for i, k in zip(input_groups, kernel_groups)
]
conv = tf.concat(3, output_groups)
return tf.reshape(tf.nn.bias_add(conv, biases),
[-1] + conv.get_shape().as_list()[1:])
x = tf.placeholder(tf.float32, (None, ) + xdim, name='imgTensor')
#conv1
#conv(11, 11, 96, 4, 4, padding='VALID', name='conv1')
k_h = 11
k_w = 11
c_o = 96
s_h = 4
s_w = 4
conv1W = tf.Variable(net_data["conv1"][0])
conv1b = tf.Variable(net_data["conv1"][1])
conv1_in = conv(x,
conv1W,
conv1b,
k_h,
k_w,
c_o,
s_h,
s_w,
padding="SAME",
group=1)
conv1 = tf.nn.relu(conv1_in)
#lrn1
#lrn(2, 2e-05, 0.75, name='norm1')
radius = 2
alpha = 2e-05
beta = 0.75
bias = 1.0
lrn1 = tf.nn.local_response_normalization(conv1,
depth_radius=radius,
alpha=alpha,
beta=beta,
bias=bias)
#maxpool1
#max_pool(3, 3, 2, 2, padding='VALID', name='pool1')
k_h = 3
k_w = 3
s_h = 2
s_w = 2
padding = 'VALID'
maxpool1 = tf.nn.max_pool(lrn1,
ksize=[1, k_h, k_w, 1],
strides=[1, s_h, s_w, 1],
padding=padding)
#conv2
#conv(5, 5, 256, 1, 1, group=2, name='conv2')
k_h = 5
k_w = 5
c_o = 256
s_h = 1
s_w = 1
group = 2
conv2W = tf.Variable(net_data["conv2"][0])
conv2b = tf.Variable(net_data["conv2"][1])
conv2_in = conv(maxpool1,
conv2W,
conv2b,
k_h,
k_w,
c_o,
s_h,
s_w,
padding="SAME",
group=group)
conv2 = tf.nn.relu(conv2_in)
#lrn2
#lrn(2, 2e-05, 0.75, name='norm2')
radius = 2
alpha = 2e-05
beta = 0.75
bias = 1.0
lrn2 = tf.nn.local_response_normalization(conv2,
depth_radius=radius,
alpha=alpha,
beta=beta,
bias=bias)
#maxpool2
#max_pool(3, 3, 2, 2, padding='VALID', name='pool2')
k_h = 3
k_w = 3
s_h = 2
s_w = 2
padding = 'VALID'
maxpool2 = tf.nn.max_pool(lrn2,
ksize=[1, k_h, k_w, 1],
strides=[1, s_h, s_w, 1],
padding=padding)
#conv3
#conv(3, 3, 384, 1, 1, name='conv3')
k_h = 3
k_w = 3
c_o = 384
s_h = 1
s_w = 1
group = 1
conv3W = tf.Variable(net_data["conv3"][0])
conv3b = tf.Variable(net_data["conv3"][1])
conv3_in = conv(maxpool2,
conv3W,
conv3b,
k_h,
k_w,
c_o,
s_h,
s_w,
padding="SAME",
group=group)
conv3 = tf.nn.relu(conv3_in)
#conv4
#conv(3, 3, 384, 1, 1, group=2, name='conv4')
k_h = 3
k_w = 3
c_o = 384
s_h = 1
s_w = 1
group = 2
conv4W = tf.Variable(net_data["conv4"][0])
conv4b = tf.Variable(net_data["conv4"][1])
conv4_in = conv(conv3,
conv4W,
conv4b,
k_h,
k_w,
c_o,
s_h,
s_w,
padding="SAME",
group=group)
conv4 = tf.nn.relu(conv4_in)
#conv5
#conv(3, 3, 256, 1, 1, group=2, name='conv5')
k_h = 3
k_w = 3
c_o = 256
s_h = 1
s_w = 1
group = 2
conv5W = tf.Variable(net_data["conv5"][0])
conv5b = tf.Variable(net_data["conv5"][1])
conv5_in = conv(conv4,
conv5W,
conv5b,
k_h,
k_w,
c_o,
s_h,
s_w,
padding="SAME",
group=group)
conv5 = tf.nn.relu(conv5_in)
#maxpool5
#max_pool(3, 3, 2, 2, padding='VALID', name='pool5')
k_h = 3
k_w = 3
s_h = 2
s_w = 2
padding = 'VALID'
maxpool5 = tf.nn.max_pool(conv5,
ksize=[1, k_h, k_w, 1],
strides=[1, s_h, s_w, 1],
padding=padding)
# maxpool5 = tf.stop_gradient(maxpool5)
#fc6
#fc(4096, name='fc6')
fc6W = tf.Variable(net_data["fc6"][0])
fc6b = tf.Variable(net_data["fc6"][1])
fc6 = tf.nn.relu_layer(tf.reshape(
maxpool5, [-1, int(prod(maxpool5.get_shape()[1:]))]),
fc6W,
fc6b,
name='fc6')
#fc7
#fc(4096, name='fc7')
fc7W = tf.Variable(net_data["fc7"][0], name='fc7W')
fc7b = tf.Variable(net_data["fc7"][1], name='fc7b')
fc7 = tf.nn.relu_layer(fc6, fc7W, fc7b, name='fc7')
fc7 = tf.stop_gradient(fc7)
W_fc2 = weight_variable([4096, self.class_count])
b_fc2 = bias_variable([self.class_count])
scores = tf.matmul(fc7, W_fc2) + b_fc2
object_or_not = tf.placeholder("float", [None])
labelTensor = tf.placeholder("float", [None, self.class_count])
return object_or_not, labelTensor, x, scores, fc7
def build_basic_graph(self, sess):
# W_conv1 = weight_variable([8, 8, 3, 32])
# b_conv1 = bias_variable([32])
# W_conv2 = weight_variable([4, 4, 32, 64])
# b_conv2 = bias_variable([64])
# W_conv3 = weight_variable([3, 3, 64, 64])
# b_conv3 = bias_variable([64])
# W_fc1 = weight_variable([6400, 512])
# b_fc1 = bias_variable([512])
# W_fc2 = weight_variable([512, self.class_count])
# b_fc2 = bias_variable([self.class_count])
# imgTensor = tf.placeholder("float", [None, 80, 80, 3])
# labelTensor = tf.placeholder("float", [None, self.class_count])
# object_or_not = tf.placeholder("float", [None])
# # hidden layers
# conv1 = tf.nn.relu(conv2d(imgTensor, W_conv1, 4) + b_conv1)
# conv2 = tf.nn.relu(conv2d(conv1, W_conv2, 2) + b_conv2)
# conv3 = tf.nn.relu(conv2d(conv2, W_conv3, 1) + b_conv3)
# conv3_flat = tf.reshape(conv3, [-1, 6400])
# h_fc1 = tf.nn.relu(tf.matmul(conv3_flat, W_fc1) + b_fc1)
# scores = tf.matmul(h_fc1, W_fc2) + b_fc2
# newSaver = tf.train.import_meta_graph(ALEXNET_META)
# newSaver.restore(sess, ALEXNET_CKPT)
# # print [n.name for n in tf.get_default_graph().as_graph_def().node]
# h_fc1 = sess.graph.get_tensor_by_name('fc7:0')
# print h_fc1
# object_or_not = tf.placeholder("float", [None])
# labelTensor = tf.placeholder(tf.float32, [None, self.class_count])
# imgTensor = sess.graph.get_tensor_by_name('imgTensor:0')
# print imgTensor
# W_fc2 = weight_variable([4096, self.class_count])
# b_fc2 = bias_variable([self.class_count])
# scores = tf.matmul(h_fc1, W_fc2) + b_fc2
train_x = np.zeros((1, 227, 227, 3)).astype(float32)
train_y = np.zeros((1, 1000))
xdim = train_x.shape[1:]
ydim = train_y.shape[1]
net_data = load("bvlc_alexnet.npy").item()
def conv(input,
kernel,
biases,
k_h,
k_w,
c_o,
s_h,
s_w,
padding="VALID",
group=1):
'''From https://github.com/ethereon/caffe-tensorflow
'''
c_i = input.get_shape()[-1]
assert c_i % group == 0
assert c_o % group == 0
convolve = lambda i, k: tf.nn.conv2d(
i, k, [1, s_h, s_w, 1], padding=padding)
if group == 1:
conv = convolve(input, kernel)
else:
input_groups = tf.split(3, group, input)
kernel_groups = tf.split(3, group, kernel)
output_groups = [
convolve(i, k) for i, k in zip(input_groups, kernel_groups)
]
conv = tf.concat(3, output_groups)
return tf.reshape(tf.nn.bias_add(conv, biases),
[-1] + conv.get_shape().as_list()[1:])
x = tf.placeholder(tf.float32, (None, ) + xdim, name='imgTensor')
#conv1
#conv(11, 11, 96, 4, 4, padding='VALID', name='conv1')
k_h = 11
k_w = 11
c_o = 96
s_h = 4
s_w = 4
conv1W = tf.Variable(net_data["conv1"][0])
conv1b = tf.Variable(net_data["conv1"][1])
conv1_in = conv(x,
conv1W,
conv1b,
k_h,
k_w,
c_o,
s_h,
s_w,
padding="SAME",
group=1)
conv1 = tf.nn.relu(conv1_in)
#lrn1
#lrn(2, 2e-05, 0.75, name='norm1')
radius = 2
alpha = 2e-05
beta = 0.75
bias = 1.0
lrn1 = tf.nn.local_response_normalization(conv1,
depth_radius=radius,
alpha=alpha,
beta=beta,
bias=bias)
#maxpool1
#max_pool(3, 3, 2, 2, padding='VALID', name='pool1')
k_h = 3
k_w = 3
s_h = 2
s_w = 2
padding = 'VALID'
maxpool1 = tf.nn.max_pool(lrn1,
ksize=[1, k_h, k_w, 1],
strides=[1, s_h, s_w, 1],
padding=padding)
#conv2
#conv(5, 5, 256, 1, 1, group=2, name='conv2')
k_h = 5
k_w = 5
c_o = 256
s_h = 1
s_w = 1
group = 2
conv2W = tf.Variable(net_data["conv2"][0])
conv2b = tf.Variable(net_data["conv2"][1])
conv2_in = conv(maxpool1,
conv2W,
conv2b,
k_h,
k_w,
c_o,
s_h,
s_w,
padding="SAME",
group=group)
conv2 = tf.nn.relu(conv2_in)
#lrn2
#lrn(2, 2e-05, 0.75, name='norm2')
radius = 2
alpha = 2e-05
beta = 0.75
bias = 1.0
lrn2 = tf.nn.local_response_normalization(conv2,
depth_radius=radius,
alpha=alpha,
beta=beta,
bias=bias)
#maxpool2
#max_pool(3, 3, 2, 2, padding='VALID', name='pool2')
k_h = 3
k_w = 3
s_h = 2
s_w = 2
padding = 'VALID'
maxpool2 = tf.nn.max_pool(lrn2,
ksize=[1, k_h, k_w, 1],
strides=[1, s_h, s_w, 1],
padding=padding)
#conv3
#conv(3, 3, 384, 1, 1, name='conv3')
k_h = 3
k_w = 3
c_o = 384
s_h = 1
s_w = 1
group = 1
conv3W = tf.Variable(net_data["conv3"][0])
conv3b = tf.Variable(net_data["conv3"][1])
conv3_in = conv(maxpool2,
conv3W,
conv3b,
k_h,
k_w,
c_o,
s_h,
s_w,
padding="SAME",
group=group)
conv3 = tf.nn.relu(conv3_in)
#conv4
#conv(3, 3, 384, 1, 1, group=2, name='conv4')
k_h = 3
k_w = 3
c_o = 384
s_h = 1
s_w = 1
group = 2
conv4W = tf.Variable(net_data["conv4"][0])
conv4b = tf.Variable(net_data["conv4"][1])
conv4_in = conv(conv3,
conv4W,
conv4b,
k_h,
k_w,
c_o,
s_h,
s_w,
padding="SAME",
group=group)
conv4 = tf.nn.relu(conv4_in)
conv4 = tf.stop_gradient(conv4)
#conv5
#conv(3, 3, 256, 1, 1, group=2, name='conv5')
k_h = 3
k_w = 3
c_o = 256
s_h = 1
s_w = 1
group = 2
conv5W = tf.Variable(net_data["conv5"][0])
conv5b = tf.Variable(net_data["conv5"][1])
conv5_in = conv(conv4,
conv5W,
conv5b,
k_h,
k_w,
c_o,
s_h,
s_w,
padding="SAME",
group=group)
conv5 = tf.nn.relu(conv5_in)
#maxpool5
#max_pool(3, 3, 2, 2, padding='VALID', name='pool5')
k_h = 3
k_w = 3
s_h = 2
s_w = 2
padding = 'VALID'
maxpool5 = tf.nn.max_pool(conv5,
ksize=[1, k_h, k_w, 1],
strides=[1, s_h, s_w, 1],
padding=padding)
# maxpool5 = tf.stop_gradient(maxpool5)
#fc6
#fc(4096, name='fc6')
fc6W = tf.Variable(net_data["fc6"][0])
fc6b = tf.Variable(net_data["fc6"][1])
fc6 = tf.nn.relu_layer(tf.reshape(
maxpool5, [-1, int(prod(maxpool5.get_shape()[1:]))]),
fc6W,
fc6b,
name='fc6')
#fc7
#fc(4096, name='fc7')
fc7W = tf.Variable(net_data["fc7"][0], name='fc7W')
fc7b = tf.Variable(net_data["fc7"][1], name='fc7b')
fc7 = tf.nn.relu_layer(fc6, fc7W, fc7b, name='fc7')
W_fc2 = weight_variable([4096, self.class_count])
b_fc2 = bias_variable([self.class_count])
scores = tf.matmul(fc7, W_fc2) + b_fc2
object_or_not = tf.placeholder("float", [None])
labelTensor = tf.placeholder("float", [None, self.class_count])
# init = tf.initialize_all_variables()
# sess = tf.Session()
# sess.run(init)
return object_or_not, labelTensor, x, scores, fc7