def inference_conv(image):
# incomplete :/
image_reshaped = tf.reshape(image, [-1, IMAGE_SIZE, IMAGE_SIZE, 1])
with tf.name_scope("conv1") as scope:
W_conv1 = utils.weight_variable([3, 3, 1, 32], name="W_conv1")
b_conv1 = utils.bias_variable([32], name="b_conv1")
add_to_reg_loss_and_summary(W_conv1, b_conv1)
h_conv1 = tf.nn.tanh(
utils.conv2d_basic(image_reshaped, W_conv1, b_conv1))
with tf.name_scope("conv2") as scope:
W_conv2 = utils.weight_variable([3, 3, 32, 64], name="W_conv2")
b_conv2 = utils.bias_variable([64], name="b_conv2")
add_to_reg_loss_and_summary(W_conv2, b_conv2)
h_conv2 = tf.nn.tanh(utils.conv2d_strided(h_conv1, W_conv2, b_conv2))
with tf.name_scope("conv3") as scope:
W_conv3 = utils.weight_variable([3, 3, 64, 128], name="W_conv3")
b_conv3 = utils.bias_variable([128], name="b_conv3")
add_to_reg_loss_and_summary(W_conv3, b_conv3)
h_conv3 = tf.nn.tanh(utils.conv2d_strided(h_conv2, W_conv3, b_conv3))
with tf.name_scope("conv4") as scope:
W_conv4 = utils.weight_variable([3, 3, 128, 256], name="W_conv4")
b_conv4 = utils.bias_variable([256], name="b_conv4")
add_to_reg_loss_and_summary(W_conv4, b_conv4)
h_conv4 = tf.nn.tanh(utils.conv2d_strided(h_conv3, W_conv4, b_conv4))
def inference(data):
with tf.variable_scope("inference") as scope:
W_1 = utils.weight_variable([IMAGE_SIZE * IMAGE_SIZE * 50], name="W_1")
b_1 = utils.bias_variable([50], name="b_1")
h_1 = tf.nn.relu(tf.matmul(data, tf.reshape(W_1, [IMAGE_SIZE * IMAGE_SIZE, 50])) + b_1, name='h_1')
utils.add_activation_summary(h_1)
W_2 = utils.weight_variable([50 * 50], name="W_2")
b_2 = utils.bias_variable([50], name="b_2")
h_2 = tf.nn.relu(tf.matmul(h_1, tf.reshape(W_2, [50, 50])) + b_2, name='h_2')
utils.add_activation_summary(h_2)
W_3 = utils.weight_variable([50 * 50], name="W_3")
b_3 = utils.bias_variable([50], name="b_3")
h_3 = tf.nn.relu(tf.matmul(h_2, tf.reshape(W_3, [50, 50])) + b_3, name='h_3')
utils.add_activation_summary(h_3)
W_4 = utils.weight_variable([50 * 50], name="W_4")
b_4 = utils.bias_variable([50], name="b_4")
h_4 = tf.nn.relu(tf.matmul(h_3, tf.reshape(W_4, [50, 50])) + b_4, name='h_4')
utils.add_activation_summary(h_4)
W_final = utils.weight_variable([50 * 10], name="W_final")
b_final = utils.bias_variable([10], name="b_final")
pred = tf.nn.softmax(tf.matmul(h_4, tf.reshape(W_final, [50, 10])) + b_final, name='h_final')
# utils.add_activation_summary(pred)
return pred
def inference_conv(image):
# incomplete :/
image_reshaped = tf.reshape(image, [-1, IMAGE_SIZE, IMAGE_SIZE, 1])
with tf.name_scope("conv1") as scope:
W_conv1 = utils.weight_variable([3, 3, 1, 32], name="W_conv1")
b_conv1 = utils.bias_variable([32], name="b_conv1")
add_to_reg_loss_and_summary(W_conv1, b_conv1)
h_conv1 = tf.nn.tanh(utils.conv2d_basic(image_reshaped, W_conv1, b_conv1))
with tf.name_scope("conv2") as scope:
W_conv2 = utils.weight_variable([3, 3, 32, 64], name="W_conv2")
b_conv2 = utils.bias_variable([64], name="b_conv2")
add_to_reg_loss_and_summary(W_conv2, b_conv2)
h_conv2 = tf.nn.tanh(utils.conv2d_strided(h_conv1, W_conv2, b_conv2))
with tf.name_scope("conv3") as scope:
W_conv3 = utils.weight_variable([3, 3, 64, 128], name="W_conv3")
b_conv3 = utils.bias_variable([128], name="b_conv3")
add_to_reg_loss_and_summary(W_conv3, b_conv3)
h_conv3 = tf.nn.tanh(utils.conv2d_strided(h_conv2, W_conv3, b_conv3))
with tf.name_scope("conv4") as scope:
W_conv4 = utils.weight_variable([3, 3, 128, 256], name="W_conv4")
b_conv4 = utils.bias_variable([256], name="b_conv4")
add_to_reg_loss_and_summary(W_conv4, b_conv4)
h_conv4 = tf.nn.tanh(utils.conv2d_strided(h_conv3, W_conv4, b_conv4))
def encode_decode(image, keep_prob):
with tf.variable_scope("encode_decode"):
#conv1
with tf.variable_scope('conv1'):
W_conv1 = weight_variable([3, 3, 3, 16])
b_conv1 = bias_variable([16])
h_conv1 = tf.nn.relu(conv2d(image, W_conv1) + b_conv1)
h_pool1 = max_pool_2x2(h_conv1)
# conv2
with tf.variable_scope('conv2'):
W_conv2 = weight_variable([3, 3, 16, 32])
b_conv2 = bias_variable([32])
h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
h_pool2 = max_pool_2x2(max_pool_2x2(h_conv2))
# conv3
with tf.variable_scope('conv3'):
W_conv3 = weight_variable([3, 3, 32, 64])
b_conv3 = bias_variable([64])
h_conv3 = tf.nn.relu(conv2d(h_pool2, W_conv3) + b_conv3)
h_pool3 = max_pool_2x2(h_conv3)
# conv4
with tf.variable_scope('conv4'):
W_conv4 = weight_variable([3, 3, 64, 128])
b_conv4 = bias_variable([128])
h_conv4 = tf.nn.relu(conv2d(h_pool3, W_conv4) + b_conv4)
h_pool4 = max_pool_2x2(h_conv4)
#Upscale
with tf.variable_scope('deconv1'):
deconv_shape1 = h_pool3.get_shape()
W_t1 = weight_variable([3, 3, deconv_shape1[3].value, 128])
b_t1 = utils.bias_variable([deconv_shape1[3].value], name="b_t1")
conv_t1 = (utils.conv2d_transpose_strided(
h_pool4, W_t1, b_t1, output_shape=tf.shape(h_pool3)))
fuse_1 = (tf.add(conv_t1, h_pool3, name="fuse_1"))
with tf.variable_scope('deconv2'):
deconv_shape2 = h_pool2.get_shape()
W_t2 = weight_variable(
[3, 3, deconv_shape2[3].value, deconv_shape1[3].value])
b_t2 = utils.bias_variable([deconv_shape2[3].value], name="b_t2")
conv_t2 = (utils.conv2d_transpose_strided(
fuse_1, W_t2, b_t2, output_shape=tf.shape(h_pool2)))
fuse_2 = (tf.add(conv_t2, h_pool2, name="fuse_2"))
with tf.variable_scope('deconv3'):
shape = tf.shape(image)
deconv_shape3 = tf.stack(
[shape[0], shape[1], shape[2], OUTPUT_CHANNELS])
W_t3 = weight_variable(
[16, 16, OUTPUT_CHANNELS, deconv_shape2[3].value])
b_t3 = utils.bias_variable([OUTPUT_CHANNELS], name="b_t3")
conv_t3 = (utils.conv2d_transpose_strided(
fuse_2, W_t3, b_t3, output_shape=deconv_shape3, stride=8))
return conv_t3
def inference_simple(dataset):
with tf.name_scope("conv1") as scope:
W1 = utils.weight_variable([5, 5, 1, 32], name="W1")
b1 = utils.bias_variable([32], name="b1")
tf.histogram_summary("W1", W1)
tf.histogram_summary("b1", b1)
h_conv1 = tf.nn.relu(utils.conv2d_basic(dataset, W1, b1), name="h_conv1")
h_pool1 = utils.max_pool_2x2(h_conv1)
with tf.name_scope("conv2") as scope:
W2 = utils.weight_variable([3, 3, 32, 64], name="W2")
b2 = utils.bias_variable([64], name="b2")
tf.histogram_summary("W2", W2)
tf.histogram_summary("b2", b2)
h_conv2 = tf.nn.relu(utils.conv2d_basic(h_pool1, W2, b2), name="h_conv2")
h_pool2 = utils.max_pool_2x2(h_conv2)
with tf.name_scope("fc") as scope:
image_size = IMAGE_SIZE // 4
h_flat = tf.reshape(h_pool2, [-1, image_size * image_size * 64])
W_fc = utils.weight_variable([image_size * image_size * 64, NUM_LABELS], name="W_fc")
b_fc = utils.bias_variable([NUM_LABELS], name="b_fc")
tf.histogram_summary("W_fc", W_fc)
tf.histogram_summary("b_fc", b_fc)
pred = tf.matmul(h_flat, W_fc) + b_fc
return pred
def up_layer(x1, x2, in_size1, in_size2, out_size, i):
# Up 1
W1 = utils.weight_variable([2, 2, 2, in_size2, in_size1],
name="W_u_" + str(i) + "_1")
b1 = utils.bias_variable([in_size2], name="b_u_" + str(i) + "_1")
deco1 = utils.conv3d_transpose_strided(x1,
W1,
b1,
output_shape=tf.shape(x2))
relu1 = tf.nn.relu(deco1, name="relu_d_" + str(i) + "_1")
# Concat
conc1 = tf.concat([relu1, x2],
-1) # concat along the channels dimension
# Conv1
W2 = utils.weight_variable([3, 3, 3, in_size2 * 2, out_size],
name="W_u_" + str(i) + "_2")
b2 = utils.bias_variable([out_size], name="b_u_" + str(i) + "_2")
conv1 = utils.conv3d_basic(conc1, W2, b2)
relu2 = tf.nn.relu(conv1, name="relu_u_" + str(i) + "_2")
relu2 = tf.nn.dropout(relu2, keep_prob=keep_prob)
# Conv2
W3 = utils.weight_variable([3, 3, 3, out_size, out_size],
name="W_u_" + str(i) + "_3")
b3 = utils.bias_variable([out_size], name="b_u_" + str(i) + "_3")
conv3 = utils.conv3d_basic(relu2, W3, b3)
relu3 = tf.nn.relu(conv3, name="relu_u_" + str(i) + "_3")
relu3 = tf.nn.dropout(relu3, keep_prob=keep_prob)
return relu3
def encoder_conv(image):
with tf.name_scope("enc_conv1") as scope:
W_conv1 = utils.weight_variable([3, 3, 3, 32], name="W_conv1")
b_conv1 = utils.bias_variable([32], name="b_conv1")
h_conv1 = tf.nn.tanh(utils.conv2d_strided(image, W_conv1, b_conv1))
with tf.name_scope("enc_conv2") as scope:
W_conv2 = utils.weight_variable([3, 3, 32, 64], name="W_conv2")
b_conv2 = utils.bias_variable([64], name="b_conv2")
h_conv2 = tf.nn.tanh(utils.conv2d_strided(h_conv1, W_conv2, b_conv2))
with tf.name_scope("enc_conv3") as scope:
W_conv3 = utils.weight_variable([3, 3, 64, 128], name="W_conv3")
b_conv3 = utils.bias_variable([128], name="b_conv3")
h_conv3 = tf.nn.tanh(utils.conv2d_strided(h_conv2, W_conv3, b_conv3))
with tf.name_scope("enc_conv4") as scope:
W_conv4 = utils.weight_variable([3, 3, 128, 256], name="W_conv4")
b_conv4 = utils.bias_variable([256], name="b_conv4")
h_conv4 = tf.nn.tanh(utils.conv2d_strided(h_conv3, W_conv4, b_conv4))
with tf.name_scope("enc_fc") as scope:
image_size = IMAGE_SIZE // 16
h_conv4_flatten = tf.reshape(h_conv4, [-1, image_size * image_size * 256])
W_fc5 = utils.weight_variable([image_size * image_size * 256, 512], name="W_fc5")
b_fc5 = utils.bias_variable([512], name="b_fc5")
encoder_val = tf.matmul(h_conv4_flatten, W_fc5) + b_fc5
return encoder_val
def inference_resnet(dataset):
dataset_reshaped = tf.reshape(dataset, [-1, 28, 28, 1])
with tf.name_scope("conv1") as scope:
W_conv1 = utils.weight_variable([5, 5, 1, 32], name="W_conv1")
bias1 = utils.bias_variable([32], name="bias1")
tf.histogram_summary("W_conv1", W_conv1)
tf.histogram_summary("bias1", bias1)
h_conv1 = tf.nn.relu(
utils.conv2d_basic(dataset_reshaped, W_conv1, bias1))
h_norm1 = utils.local_response_norm(h_conv1)
bottleneck_1 = utils.bottleneck_unit(h_norm1,
32,
32,
down_stride=True,
name="res1")
bottleneck_2 = utils.bottleneck_unit(bottleneck_1,
64,
64,
down_stride=True,
name="res2")
with tf.name_scope("fc1") as scope:
h_flat = tf.reshape(bottleneck_2, [-1, 7 * 7 * 64])
W_fc1 = utils.weight_variable([7 * 7 * 64, 10], name="W_fc1")
bias_fc1 = utils.bias_variable([10], name="bias_fc1")
tf.histogram_summary("W_fc1", W_fc1)
tf.histogram_summary("bias_fc1", bias_fc1)
logits = tf.matmul(h_flat, W_fc1) + bias_fc1
return logits
def inference_conv(dataset):
dataset_reshaped = tf.reshape(dataset, [-1, 28, 28, 1])
with tf.name_scope("conv1") as scope:
W_conv1 = utils.weight_variable([5, 5, 1, 32], name="W_conv1")
bias1 = utils.bias_variable([32], name="bias1")
tf.histogram_summary("W_conv1", W_conv1)
tf.histogram_summary("bias1", bias1)
h_conv1 = tf.nn.relu(
utils.conv2d_basic(dataset_reshaped, W_conv1, bias1))
h_norm1 = utils.local_response_norm(h_conv1)
h_pool1 = utils.max_pool_2x2(h_norm1)
with tf.name_scope("conv2") as scope:
W_conv2 = utils.weight_variable([3, 3, 32, 64], name="W_conv2")
bias2 = utils.bias_variable([64], name="bias2")
tf.histogram_summary("W_conv2", W_conv2)
tf.histogram_summary("bias2", bias2)
h_conv2 = tf.nn.relu(utils.conv2d_basic(h_pool1, W_conv2, bias2))
h_norm2 = utils.local_response_norm(h_conv2)
h_pool2 = utils.max_pool_2x2(h_norm2)
with tf.name_scope("fc1") as scope:
h_flat = tf.reshape(h_pool2, [-1, 7 * 7 * 64])
W_fc1 = utils.weight_variable([7 * 7 * 64, 10], name="W_fc1")
bias_fc1 = utils.bias_variable([10], name="bias_fc1")
tf.histogram_summary("W_fc1", W_fc1)
tf.histogram_summary("bias_fc1", bias_fc1)
logits = tf.matmul(h_flat, W_fc1) + bias_fc1
return logits
def encoder_conv(image):
with tf.name_scope("enc_conv1") as scope:
W_conv1 = utils.weight_variable([3, 3, 3, 32], name="W_conv1")
b_conv1 = utils.bias_variable([32], name="b_conv1")
h_conv1 = tf.nn.tanh(utils.conv2d_strided(image, W_conv1, b_conv1))
with tf.name_scope("enc_conv2") as scope:
W_conv2 = utils.weight_variable([3, 3, 32, 64], name="W_conv2")
b_conv2 = utils.bias_variable([64], name="b_conv2")
h_conv2 = tf.nn.tanh(utils.conv2d_strided(h_conv1, W_conv2, b_conv2))
with tf.name_scope("enc_conv3") as scope:
W_conv3 = utils.weight_variable([3, 3, 64, 128], name="W_conv3")
b_conv3 = utils.bias_variable([128], name="b_conv3")
h_conv3 = tf.nn.tanh(utils.conv2d_strided(h_conv2, W_conv3, b_conv3))
with tf.name_scope("enc_conv4") as scope:
W_conv4 = utils.weight_variable([3, 3, 128, 256], name="W_conv4")
b_conv4 = utils.bias_variable([256], name="b_conv4")
h_conv4 = tf.nn.tanh(utils.conv2d_strided(h_conv3, W_conv4, b_conv4))
with tf.name_scope("enc_fc") as scope:
image_size = IMAGE_SIZE // 16
h_conv4_flatten = tf.reshape(h_conv4,
[-1, image_size * image_size * 256])
W_fc5 = utils.weight_variable([image_size * image_size * 256, 512],
name="W_fc5")
b_fc5 = utils.bias_variable([512], name="b_fc5")
encoder_val = tf.matmul(h_conv4_flatten, W_fc5) + b_fc5
return encoder_val
def inference(data):
with tf.variable_scope("inference") as scope:
W_1 = utils.weight_variable([IMAGE_SIZE * IMAGE_SIZE * 50], name="W_1")
b_1 = utils.bias_variable([50], name="b_1")
h_1 = tf.nn.relu(tf.matmul(data, tf.reshape(W_1, [IMAGE_SIZE * IMAGE_SIZE, 50])) + b_1, name='h_1')
utils.add_activation_summary(h_1)
W_2 = utils.weight_variable([50 * 50], name="W_2")
b_2 = utils.bias_variable([50], name="b_2")
h_2 = tf.nn.relu(tf.matmul(h_1, tf.reshape(W_2, [50, 50])) + b_2, name='h_2')
utils.add_activation_summary(h_2)
W_3 = utils.weight_variable([50 * 50], name="W_3")
b_3 = utils.bias_variable([50], name="b_3")
h_3 = tf.nn.relu(tf.matmul(h_2, tf.reshape(W_3, [50, 50])) + b_3, name='h_3')
utils.add_activation_summary(h_3)
W_4 = utils.weight_variable([50 * 50], name="W_4")
b_4 = utils.bias_variable([50], name="b_4")
h_4 = tf.nn.relu(tf.matmul(h_3, tf.reshape(W_4, [50, 50])) + b_4, name='h_4')
utils.add_activation_summary(h_4)
W_final = utils.weight_variable([50 * 10], name="W_final")
b_final = utils.bias_variable([10], name="b_final")
pred = tf.nn.softmax(tf.matmul(h_4, tf.reshape(W_final, [50, 10])) + b_final, name='h_final')
# utils.add_activation_summary(pred)
return pred
def upsample(pool5, pool4, pool3, pool2, conv9, image, scope, output_class):
with tf.variable_scope(scope):
# do the upscaling using 2 fuse layers
deconv_shape1 = pool5.get_shape()
W_t1 = utils.weight_variable(
[4, 4, deconv_shape1[3].value, NUM_OF_CLASSESS], name="W_t1")
b_t1 = utils.bias_variable([deconv_shape1[3].value], name="b_t1")
conv_t1 = utils.conv2d_transpose_strided(conv9,
W_t1,
b_t1,
output_shape=tf.shape(pool5))
fuse_1 = tf.add(conv_t1, pool5, name="fuse_1")
deconv_shape2 = pool4.get_shape()
W_t2 = utils.weight_variable(
[4, 4, deconv_shape2[3].value, deconv_shape1[3].value],
name="W_t2")
b_t2 = utils.bias_variable([deconv_shape2[3].value], name="b_t2")
conv_t2 = utils.conv2d_transpose_strided(fuse_1,
W_t2,
b_t2,
output_shape=tf.shape(pool4))
fuse_2 = tf.add(conv_t2, pool4, name="fuse_2")
deconv_shape3 = pool3.get_shape()
W_t3 = utils.weight_variable(
[4, 4, deconv_shape3[3].value, deconv_shape2[3].value],
name="W_t3")
b_t3 = utils.bias_variable([deconv_shape3[3].value], name="b_t3")
conv_t3 = utils.conv2d_transpose_strided(fuse_2,
W_t3,
b_t3,
output_shape=tf.shape(pool3))
fuse_3 = tf.add(conv_t3, pool3, name="fuse_3")
deconv_shape4 = pool2.get_shape()
W_t4 = utils.weight_variable(
[4, 4, deconv_shape4[3].value, deconv_shape3[3].value],
name="W_t4")
b_t4 = utils.bias_variable([deconv_shape4[3].value], name="b_t4")
conv_t4 = utils.conv2d_transpose_strided(fuse_3,
W_t4,
b_t4,
output_shape=tf.shape(pool2))
fuse_4 = tf.add(conv_t4, pool2, name="fuse_4")
# do the final upscaling
shape = tf.shape(image)
deconv_shape5 = tf.stack([shape[0], shape[1], shape[2], output_class])
W_t5 = utils.weight_variable(
[16, 16, output_class, deconv_shape4[3].value], name="W_t5")
b_t5 = utils.bias_variable([output_class], name="b_t5")
conv_t5 = utils.conv2d_transpose_strided(fuse_4,
W_t5,
b_t5,
output_shape=deconv_shape5,
stride=2)
return conv_t5
def generator(images, train_phase):
print("setting up vgg initialized conv layers ...")
model_data = utils.get_model_data(FLAGS.model_dir, MODEL_URL)
weights = np.squeeze(model_data['layers'])
with tf.variable_scope("generator") as scope:
W0 = utils.weight_variable([3, 3, 1, 64], name="W0")
b0 = utils.bias_variable([64], name="b0")
conv0 = utils.conv2d_basic(images, W0, b0)
hrelu0 = tf.nn.relu(conv0, name="relu")
image_net = vgg_net(weights, hrelu0)
vgg_final_layer = image_net["relu5_3"]
pool5 = utils.max_pool_2x2(vgg_final_layer)
# now to upscale to actual image size
deconv_shape1 = image_net["pool4"].get_shape()
W_t1 = utils.weight_variable(
[4, 4, deconv_shape1[3].value,
pool5.get_shape()[3].value],
name="W_t1")
b_t1 = utils.bias_variable([deconv_shape1[3].value], name="b_t1")
conv_t1 = utils.conv2d_transpose_strided(pool5,
W_t1,
b_t1,
output_shape=tf.shape(
image_net["pool4"]))
fuse_1 = tf.add(conv_t1, image_net["pool4"], name="fuse_1")
deconv_shape2 = image_net["pool3"].get_shape()
W_t2 = utils.weight_variable(
[4, 4, deconv_shape2[3].value, deconv_shape1[3].value],
name="W_t2")
b_t2 = utils.bias_variable([deconv_shape2[3].value], name="b_t2")
conv_t2 = utils.conv2d_transpose_strided(fuse_1,
W_t2,
b_t2,
output_shape=tf.shape(
image_net["pool3"]))
fuse_2 = tf.add(conv_t2, image_net["pool3"], name="fuse_2")
shape = tf.shape(images)
deconv_shape3 = tf.stack([shape[0], shape[1], shape[2], 2])
W_t3 = utils.weight_variable([16, 16, 2, deconv_shape2[3].value],
name="W_t3")
b_t3 = utils.bias_variable([2], name="b_t3")
pred = utils.conv2d_transpose_strided(fuse_2,
W_t3,
b_t3,
output_shape=deconv_shape3,
stride=8)
# return tf.concat(concat_dim=3, values=[images, pred], name="pred_image")
return tf.concat([images, pred], 3, "pred_image")
def decoder_conv(embedding):
image_size = IMAGE_SIZE // 16
with tf.name_scope("dec_fc") as scope:
W_fc1 = utils.weight_variable([512, image_size * image_size * 256],
name="W_fc1")
b_fc1 = utils.bias_variable([image_size * image_size * 256],
name="b_fc1")
h_fc1 = tf.nn.relu(tf.matmul(embedding, W_fc1) + b_fc1)
with tf.name_scope("dec_conv1") as scope:
h_reshaped = tf.reshape(
h_fc1, tf.pack([tf.shape(h_fc1)[0], image_size, image_size, 256]))
W_conv_t1 = utils.weight_variable([3, 3, 128, 256], name="W_conv_t1")
b_conv_t1 = utils.bias_variable([128], name="b_conv_t1")
deconv_shape = tf.pack(
[tf.shape(h_fc1)[0], 2 * image_size, 2 * image_size, 128])
h_conv_t1 = tf.nn.relu(
utils.conv2d_transpose_strided(h_reshaped,
W_conv_t1,
b_conv_t1,
output_shape=deconv_shape))
with tf.name_scope("dec_conv2") as scope:
W_conv_t2 = utils.weight_variable([3, 3, 64, 128], name="W_conv_t2")
b_conv_t2 = utils.bias_variable([64], name="b_conv_t2")
deconv_shape = tf.pack(
[tf.shape(h_conv_t1)[0], 4 * image_size, 4 * image_size, 64])
h_conv_t2 = tf.nn.relu(
utils.conv2d_transpose_strided(h_conv_t1,
W_conv_t2,
b_conv_t2,
output_shape=deconv_shape))
with tf.name_scope("dec_conv3") as scope:
W_conv_t3 = utils.weight_variable([3, 3, 32, 64], name="W_conv_t3")
b_conv_t3 = utils.bias_variable([32], name="b_conv_t3")
deconv_shape = tf.pack(
[tf.shape(h_conv_t2)[0], 8 * image_size, 8 * image_size, 32])
h_conv_t3 = tf.nn.relu(
utils.conv2d_transpose_strided(h_conv_t2,
W_conv_t3,
b_conv_t3,
output_shape=deconv_shape))
with tf.name_scope("dec_conv4") as scope:
W_conv_t4 = utils.weight_variable([3, 3, 3, 32], name="W_conv_t4")
b_conv_t4 = utils.bias_variable([3], name="b_conv_t4")
deconv_shape = tf.pack(
[tf.shape(h_conv_t3)[0], IMAGE_SIZE, IMAGE_SIZE, 3])
pred_image = utils.conv2d_transpose_strided(h_conv_t3,
W_conv_t4,
b_conv_t4,
output_shape=deconv_shape)
return pred_image
def generator(z, train_mode):
with tf.variable_scope("generator") as scope:
W_0 = utils.weight_variable([FLAGS.z_dim, 64 * GEN_DIMENSION / 2 * IMAGE_SIZE / 16 * IMAGE_SIZE / 16],
name="W_0")
b_0 = utils.bias_variable([64 * GEN_DIMENSION / 2 * IMAGE_SIZE / 16 * IMAGE_SIZE / 16], name="b_0")
z_0 = tf.matmul(z, W_0) + b_0
h_0 = tf.reshape(z_0, [-1, IMAGE_SIZE / 16, IMAGE_SIZE / 16, 64 * GEN_DIMENSION / 2])
h_bn0 = utils.batch_norm(h_0, 64 * GEN_DIMENSION / 2, train_mode, scope="gen_bn0")
h_relu0 = tf.nn.relu(h_bn0, name='relu0')
utils.add_activation_summary(h_relu0)
# W_1 = utils.weight_variable([5, 5, 64 * GEN_DIMENSION/2, 64 * GEN_DIMENSION], name="W_1")
# b_1 = utils.bias_variable([64 * GEN_DIMENSION/2], name="b_1")
# deconv_shape = tf.pack([tf.shape(h_relu0)[0], IMAGE_SIZE / 16, IMAGE_SIZE / 16, 64 * GEN_DIMENSION/2])
# h_conv_t1 = utils.conv2d_transpose_strided(h_relu0, W_1, b_1, output_shape=deconv_shape)
# h_bn1 = utils.batch_norm(h_conv_t1, 64 * GEN_DIMENSION/2, train_mode, scope="gen_bn1")
# h_relu1 = tf.nn.relu(h_bn1, name='relu1')
# utils.add_activation_summary(h_relu1)
W_2 = utils.weight_variable([5, 5, 64 * GEN_DIMENSION / 4, 64 * GEN_DIMENSION / 2],
name="W_2")
b_2 = utils.bias_variable([64 * GEN_DIMENSION / 4], name="b_2")
deconv_shape = tf.pack([tf.shape(h_relu0)[0], IMAGE_SIZE / 8, IMAGE_SIZE / 8, 64 * GEN_DIMENSION / 4])
h_conv_t2 = utils.conv2d_transpose_strided(h_relu0, W_2, b_2, output_shape=deconv_shape)
h_bn2 = utils.batch_norm(h_conv_t2, 64 * GEN_DIMENSION / 4, train_mode, scope="gen_bn2")
h_relu2 = tf.nn.relu(h_bn2, name='relu2')
utils.add_activation_summary(h_relu2)
W_3 = utils.weight_variable([5, 5, 64 * GEN_DIMENSION / 8, 64 * GEN_DIMENSION / 4],
name="W_3")
b_3 = utils.bias_variable([64 * GEN_DIMENSION / 8], name="b_3")
deconv_shape = tf.pack([tf.shape(h_relu2)[0], IMAGE_SIZE / 4, IMAGE_SIZE / 4, 64 * GEN_DIMENSION / 8])
h_conv_t3 = utils.conv2d_transpose_strided(h_relu2, W_3, b_3, output_shape=deconv_shape)
h_bn3 = utils.batch_norm(h_conv_t3, 64 * GEN_DIMENSION / 8, train_mode, scope="gen_bn3")
h_relu3 = tf.nn.relu(h_bn3, name='relu3')
utils.add_activation_summary(h_relu3)
W_4 = utils.weight_variable([5, 5, 64 * GEN_DIMENSION / 16, 64 * GEN_DIMENSION / 8],
name="W_4")
b_4 = utils.bias_variable([64 * GEN_DIMENSION / 16], name="b_4")
deconv_shape = tf.pack([tf.shape(h_relu3)[0], IMAGE_SIZE / 2, IMAGE_SIZE / 2, 64 * GEN_DIMENSION / 16])
h_conv_t4 = utils.conv2d_transpose_strided(h_relu3, W_4, b_4, output_shape=deconv_shape)
h_bn4 = utils.batch_norm(h_conv_t4, 64 * GEN_DIMENSION / 16, train_mode, scope="gen_bn4")
h_relu4 = tf.nn.relu(h_bn4, name='relu4')
utils.add_activation_summary(h_relu4)
W_5 = utils.weight_variable([5, 5, NUM_OF_CHANNELS, 64 * GEN_DIMENSION / 16], name="W_5")
b_5 = utils.bias_variable([NUM_OF_CHANNELS], name="b_5")
deconv_shape = tf.pack([tf.shape(h_relu4)[0], IMAGE_SIZE, IMAGE_SIZE, NUM_OF_CHANNELS])
h_conv_t5 = utils.conv2d_transpose_strided(h_relu4, W_5, b_5, output_shape=deconv_shape)
pred_image = tf.nn.tanh(h_conv_t5, name='pred_image')
utils.add_activation_summary(pred_image)
return pred_image
def inference(image, keep_prob):
"""
Semantic segmentation network definition
:param image: input image. Should have values in range 0-255
:param keep_prob:
:return:
"""
print("setting up inception_v3 initialized conv layers ...")
with tf.variable_scope("inference"):
net, end_points = inception_v3(image, NUM_OF_CLASSESS, True, keep_prob)
# now to upscale to actual image size
with tf.variable_scope('Upsampling'):
with slim.arg_scope([slim.conv2d_transpose],
stride=2,
padding='SAME'):
up_sampling = end_points['']
deconv_shape1 = image_net["pool4"].get_shape()
W_t1 = utils.weight_variable(
[4, 4, deconv_shape1[3].value, NUM_OF_CLASSESS], name="W_t1")
b_t1 = utils.bias_variable([deconv_shape1[3].value], name="b_t1")
conv_t1 = utils.conv2d_transpose_strided(conv8,
W_t1,
b_t1,
output_shape=tf.shape(
image_net["pool4"]))
fuse_1 = tf.add(conv_t1, image_net["pool4"], name="fuse_1")
deconv_shape2 = image_net["pool3"].get_shape()
W_t2 = utils.weight_variable(
[4, 4, deconv_shape2[3].value, deconv_shape1[3].value],
name="W_t2")
b_t2 = utils.bias_variable([deconv_shape2[3].value], name="b_t2")
conv_t2 = utils.conv2d_transpose_strided(fuse_1,
W_t2,
b_t2,
output_shape=tf.shape(
image_net["pool3"]))
fuse_2 = tf.add(conv_t2, image_net["pool3"], name="fuse_2")
shape = tf.shape(image)
deconv_shape3 = tf.stack(
[shape[0], shape[1], shape[2], NUM_OF_CLASSESS])
W_t3 = utils.weight_variable(
[16, 16, NUM_OF_CLASSESS, deconv_shape2[3].value], name="W_t3")
b_t3 = utils.bias_variable([NUM_OF_CLASSESS], name="b_t3")
conv_t3 = utils.conv2d_transpose_strided(fuse_2,
W_t3,
b_t3,
output_shape=deconv_shape3,
stride=8)
annotation_pred = tf.argmax(conv_t3, dimension=3, name="prediction")
return tf.expand_dims(annotation_pred, dim=3), conv_t3
def FCN(image, keep_prob):
with tf.variable_scope("FCN"):
#conv1
with tf.variable_scope('conv1'):
W_conv1 = weight_variable([3, 3, 3, 32])
b_conv1 = bias_variable([32])
h_conv1 = tf.nn.relu(conv2d(image, W_conv1) + b_conv1)
h_pool1 = max_pool_2x2(h_conv1)
# conv2
with tf.variable_scope('conv2'):
W_conv2 = weight_variable([3, 3, 32, 128])
b_conv2 = bias_variable([128])
h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
h_pool2 = max_pool_2x2(max_pool_2x2(h_conv2))
# conv3
with tf.variable_scope('conv3'):
W_conv3 = weight_variable([3, 3, 128, 256])
b_conv3 = bias_variable([256])
h_conv3 = tf.nn.relu(conv2d(h_pool2, W_conv3) + b_conv3)
h_pool3 = max_pool_2x2(h_conv3)
# conv4
with tf.variable_scope('conv4'):
W_conv4 = weight_variable([3, 3, 256, 512])
b_conv4 = bias_variable([512])
h_conv4 = tf.nn.relu(conv2d(h_pool3, W_conv4) + b_conv4)
h_pool4 = max_pool_2x2(h_conv4)
#Upscale
deconv_shape1 = h_pool3.get_shape()
W_t1 = utils.weight_variable([4, 4, deconv_shape1[3].value, 512], name="W_t1")
b_t1 = utils.bias_variable([deconv_shape1[3].value], name="b_t1")
conv_t1 = utils.conv2d_transpose_strided(h_pool4, W_t1, b_t1, output_shape=tf.shape(h_pool3))
fuse_1 = tf.add(conv_t1, h_pool3, name="fuse_1")
deconv_shape2 = h_pool2.get_shape()
W_t2 = utils.weight_variable([4, 4, deconv_shape2[3].value, deconv_shape1[3].value], name="W_t2")
b_t2 = utils.bias_variable([deconv_shape2[3].value], name="b_t2")
conv_t2 = utils.conv2d_transpose_strided(fuse_1, W_t2, b_t2, output_shape=tf.shape(h_pool2))
fuse_2 = tf.add(conv_t2, h_pool2, name="fuse_2")
shape = tf.shape(image)
deconv_shape3 = tf.stack([shape[0], shape[1], shape[2], NUM_OF_CLASSES_FCN])
W_t3 = utils.weight_variable([16, 16, NUM_OF_CLASSES_FCN, deconv_shape2[3].value], name="W_t3")
b_t3 = utils.bias_variable([NUM_OF_CLASSES_FCN], name="b_t3")
conv_t3 = utils.conv2d_transpose_strided(fuse_2, W_t3, b_t3, output_shape=deconv_shape3, stride=8)
annotation_pred = tf.argmax(conv_t3, dimension=3, name="prediction")
return tf.expand_dims(annotation_pred, dim=3), conv_t3
def inference(image, keep_prob):
"""
Semantic segmentation network definition
:param image: input image. Should have values in range 0-255
:param keep_prob:
:return:
"""
print("setting up vgg initialized conv layers ...")
model_data = utils.get_model_data(model_dir, MODEL_URL)
mean = model_data['normalization'][0][0][0]
mean_pixel = np.mean(mean, axis=(0, 1))
weights = np.squeeze(model_data['layers'])
processed_image = utils.process_image(image, mean_pixel)
with tf.variable_scope("inference"):
image_net = vgg_net(weights, processed_image)
conv_final_layer = image_net["conv4_3"]
W6 = utils.weight_variable([1, 1, 512, 1024], name="W6", init=weight_init)
b6 = utils.bias_variable([1024], name="b6")
conv6 = utils.conv2d_basic(conv_final_layer, W6, b6)
relu6 = tf.nn.relu(conv6, name="relu6")
if debug:
utils.add_activation_summary(relu6)
relu_dropout6 = tf.nn.dropout(relu6, keep_prob=keep_prob)
W7 = utils.weight_variable([1, 1, 1024, NUM_OF_CLASSESS], name="W7", init=weight_init)
b7 = utils.bias_variable([NUM_OF_CLASSESS], name="b7")
conv7 = utils.conv2d_basic(relu_dropout6, W7, b7)
# now to upscale to actual image size
deconv_shape1 = image_net["pool2"].get_shape()
W_t1 = utils.weight_variable([4, 4, deconv_shape1[3].value, NUM_OF_CLASSESS], name="W_t1", init=weight_init)
b_t1 = utils.bias_variable([deconv_shape1[3].value], name="b_t1")
conv_t1 = utils.conv2d_transpose_strided(conv7, W_t1, b_t1, output_shape=tf.shape(image_net["pool2"]))
fuse_1 = tf.add(conv_t1, image_net["pool2"], name="fuse_1")
deconv_shape2 = image_net["pool1"].get_shape()
W_t2 = utils.weight_variable([4, 4, deconv_shape2[3].value, deconv_shape1[3].value], name="W_t2", init=weight_init)
b_t2 = utils.bias_variable([deconv_shape2[3].value], name="b_t2")
conv_t2 = utils.conv2d_transpose_strided(fuse_1, W_t2, b_t2, output_shape=tf.shape(image_net["pool1"]))
fuse_2 = tf.add(conv_t2, image_net["pool1"], name="fuse_1")
shape = tf.shape(image)
deconv_shape3 = tf.stack([shape[0], shape[1], shape[2], NUM_OF_CLASSESS])
W_t3 = utils.weight_variable([4, 4, NUM_OF_CLASSESS, deconv_shape2[3].value], name="W_t3", init=weight_init)
b_t3 = utils.bias_variable([NUM_OF_CLASSESS], name="b_t3")
conv_t3 = utils.conv2d_transpose_strided(fuse_2, W_t3, b_t3, output_shape=deconv_shape3, stride=2)
return conv_t3
def generator(z, train_mode):
with tf.variable_scope("generator") as scope:
W_0 = utils.weight_variable([FLAGS.z_dim, 64 * GEN_DIMENSION / 2 * IMAGE_SIZE / 16 * IMAGE_SIZE / 16],
name="W_0")
b_0 = utils.bias_variable([64 * GEN_DIMENSION / 2 * IMAGE_SIZE / 16 * IMAGE_SIZE / 16], name="b_0")
z_0 = tf.matmul(z, W_0) + b_0
h_0 = tf.reshape(z_0, [-1, IMAGE_SIZE / 16, IMAGE_SIZE / 16, 64 * GEN_DIMENSION / 2])
h_bn0 = utils.batch_norm(h_0, 64 * GEN_DIMENSION / 2, train_mode, scope="gen_bn0")
h_relu0 = tf.nn.relu(h_bn0, name='relu0')
utils.add_activation_summary(h_relu0)
# W_1 = utils.weight_variable([5, 5, 64 * GEN_DIMENSION/2, 64 * GEN_DIMENSION], name="W_1")
# b_1 = utils.bias_variable([64 * GEN_DIMENSION/2], name="b_1")
# deconv_shape = tf.pack([tf.shape(h_relu0)[0], IMAGE_SIZE / 16, IMAGE_SIZE / 16, 64 * GEN_DIMENSION/2])
# h_conv_t1 = utils.conv2d_transpose_strided(h_relu0, W_1, b_1, output_shape=deconv_shape)
# h_bn1 = utils.batch_norm(h_conv_t1, 64 * GEN_DIMENSION/2, train_mode, scope="gen_bn1")
# h_relu1 = tf.nn.relu(h_bn1, name='relu1')
# utils.add_activation_summary(h_relu1)
W_2 = utils.weight_variable([5, 5, 64 * GEN_DIMENSION / 4, 64 * GEN_DIMENSION / 2],
name="W_2")
b_2 = utils.bias_variable([64 * GEN_DIMENSION / 4], name="b_2")
deconv_shape = tf.pack([tf.shape(h_relu0)[0], IMAGE_SIZE / 8, IMAGE_SIZE / 8, 64 * GEN_DIMENSION / 4])
h_conv_t2 = utils.conv2d_transpose_strided(h_relu0, W_2, b_2, output_shape=deconv_shape)
h_bn2 = utils.batch_norm(h_conv_t2, 64 * GEN_DIMENSION / 4, train_mode, scope="gen_bn2")
h_relu2 = tf.nn.relu(h_bn2, name='relu2')
utils.add_activation_summary(h_relu2)
W_3 = utils.weight_variable([5, 5, 64 * GEN_DIMENSION / 8, 64 * GEN_DIMENSION / 4],
name="W_3")
b_3 = utils.bias_variable([64 * GEN_DIMENSION / 8], name="b_3")
deconv_shape = tf.pack([tf.shape(h_relu2)[0], IMAGE_SIZE / 4, IMAGE_SIZE / 4, 64 * GEN_DIMENSION / 8])
h_conv_t3 = utils.conv2d_transpose_strided(h_relu2, W_3, b_3, output_shape=deconv_shape)
h_bn3 = utils.batch_norm(h_conv_t3, 64 * GEN_DIMENSION / 8, train_mode, scope="gen_bn3")
h_relu3 = tf.nn.relu(h_bn3, name='relu3')
utils.add_activation_summary(h_relu3)
W_4 = utils.weight_variable([5, 5, 64 * GEN_DIMENSION / 16, 64 * GEN_DIMENSION / 8],
name="W_4")
b_4 = utils.bias_variable([64 * GEN_DIMENSION / 16], name="b_4")
deconv_shape = tf.pack([tf.shape(h_relu3)[0], IMAGE_SIZE / 2, IMAGE_SIZE / 2, 64 * GEN_DIMENSION / 16])
h_conv_t4 = utils.conv2d_transpose_strided(h_relu3, W_4, b_4, output_shape=deconv_shape)
h_bn4 = utils.batch_norm(h_conv_t4, 64 * GEN_DIMENSION / 16, train_mode, scope="gen_bn4")
h_relu4 = tf.nn.relu(h_bn4, name='relu4')
utils.add_activation_summary(h_relu4)
W_5 = utils.weight_variable([5, 5, NUM_OF_CHANNELS, 64 * GEN_DIMENSION / 16], name="W_5")
b_5 = utils.bias_variable([NUM_OF_CHANNELS], name="b_5")
deconv_shape = tf.pack([tf.shape(h_relu4)[0], IMAGE_SIZE, IMAGE_SIZE, NUM_OF_CHANNELS])
h_conv_t5 = utils.conv2d_transpose_strided(h_relu4, W_5, b_5, output_shape=deconv_shape)
pred_image = tf.nn.tanh(h_conv_t5, name='pred_image')
utils.add_activation_summary(pred_image)
return pred_image
def inference(inputs):
with tf.name_scope("input"):
W1 = utils.weight_variable([2, NEURONS_PER_LAYER], name="weights_1")
b1 = utils.bias_variable([NEURONS_PER_LAYER], name="bias_1")
tf.histogram_summary("W1", W1)
tf.histogram_summary("b1", b1)
h1 = tf.nn.relu(tf.nn.bias_add(tf.matmul(inputs, W1), b1))
with tf.name_scope("hidden2"):
W2 = utils.weight_variable([NEURONS_PER_LAYER, NEURONS_PER_LAYER], name="weights_2")
b2 = utils.bias_variable([NEURONS_PER_LAYER], name="bias_2")
tf.histogram_summary("W2", W2)
tf.histogram_summary("b2", b2)
h2 = tf.nn.relu(tf.matmul(h1, W2) + b2)
with tf.name_scope("hidden3"):
W3 = utils.weight_variable([NEURONS_PER_LAYER, NEURONS_PER_LAYER], name="weights_3")
b3 = utils.bias_variable([NEURONS_PER_LAYER], name="bias_3")
tf.histogram_summary("W3", W3)
tf.histogram_summary("b3", b3)
h3 = tf.nn.relu(tf.matmul(h2, W3) + b3)
with tf.name_scope("hidden4"):
W4 = utils.weight_variable([NEURONS_PER_LAYER, NEURONS_PER_LAYER], name="weights_4")
b4 = utils.bias_variable([NEURONS_PER_LAYER], name="bias_4")
tf.histogram_summary("W4", W4)
tf.histogram_summary("b4", b4)
h4 = tf.nn.relu(tf.matmul(h3, W4) + b4)
with tf.name_scope("hidden5"):
W5 = utils.weight_variable([NEURONS_PER_LAYER, NEURONS_PER_LAYER], name="weights_5")
b5 = utils.bias_variable([NEURONS_PER_LAYER], name="bias_5")
tf.histogram_summary("W5", W5)
tf.histogram_summary("b5", b5)
h5 = tf.nn.relu(tf.matmul(h4, W5) + b5)
with tf.name_scope("hidden6"):
W6 = utils.weight_variable([NEURONS_PER_LAYER, NEURONS_PER_LAYER], name="weights_6")
b6 = utils.bias_variable([NEURONS_PER_LAYER], name="bias_6")
tf.histogram_summary("W6", W6)
tf.histogram_summary("b6", b6)
h6 = tf.nn.relu(tf.matmul(h5, W6) + b6)
with tf.name_scope("hidden7"):
W7 = utils.weight_variable([NEURONS_PER_LAYER, NEURONS_PER_LAYER], name="weights_7")
b7 = utils.bias_variable([NEURONS_PER_LAYER], name="bias_7")
tf.histogram_summary("W7", W6)
tf.histogram_summary("b7", b6)
h7 = tf.nn.relu(tf.matmul(h6, W7) + b7)
with tf.name_scope("hidden8"):
W8 = utils.weight_variable([NEURONS_PER_LAYER, NEURONS_PER_LAYER], name="weights_8")
b8 = utils.bias_variable([NEURONS_PER_LAYER], name="bias_8")
tf.histogram_summary("W8", W6)
tf.histogram_summary("b8", b6)
h8 = tf.nn.relu(tf.matmul(h7, W8) + b8)
with tf.name_scope("output"):
W9 = utils.weight_variable([NEURONS_PER_LAYER, channels], name="weights_9")
b9 = utils.bias_variable([channels], name="bias_9")
tf.histogram_summary("W9", W9)
tf.histogram_summary("b9", b9)
pred = tf.matmul(h8, W9) + b9
return pred
def inference_res(input_image):
W1 = utils.weight_variable([3, 3, 3, 32])
b1 = utils.bias_variable([32])
hconv_1 = tf.nn.relu(utils.conv2d_basic(input_image, W1, b1))
h_norm = utils.local_response_norm(hconv_1)
bottleneck_1 = utils.bottleneck_unit(h_norm, 16, 16, down_stride=True, name="res_1")
bottleneck_2 = utils.bottleneck_unit(bottleneck_1, 8, 8, down_stride=True, name="res_2")
bottleneck_3 = utils.bottleneck_unit(bottleneck_2, 16, 16, up_stride=True, name="res_3")
bottleneck_4 = utils.bottleneck_unit(bottleneck_3, 32, 32, up_stride=True, name="res_4")
W5 = utils.weight_variable([3, 3, 32, 3])
b5 = utils.bias_variable([3])
out = tf.nn.tanh(utils.conv2d_basic(bottleneck_4, W5, b5))
return out
def decoder_fc(z):
with tf.variable_scope("decoder") as scope:
Wd_fc1 = utils.weight_variable([FLAGS.z_dim, 50], name="Wd_fc1")
bd_fc1 = utils.bias_variable([50], name="bd_fc1")
hd_relu1 = activation_function(tf.matmul(z, Wd_fc1) + bd_fc1, name="hdfc_1")
Wd_fc2 = utils.weight_variable([50, 50], name="Wd_fc2")
bd_fc2 = utils.bias_variable([50], name="bd_fc2")
hd_relu2 = activation_function(tf.matmul(hd_relu1, Wd_fc2) + bd_fc2, name="hdfc_2")
Wd_fc3 = utils.weight_variable([50, IMAGE_SIZE * IMAGE_SIZE], name="Wd_fc3")
bd_fc3 = utils.bias_variable([IMAGE_SIZE * IMAGE_SIZE], name="bd_fc3")
pred_image = tf.matmul(hd_relu2, Wd_fc3) + bd_fc3
return pred_image
def deconv2d_layer_concat(x,
name,
W_s,
concat_x,
output_shape=None,
stride=2,
stddev=0.02,
if_relu=False):
'''
Deconv2d operator for U-Net concat.
Args:
x: inputs
W_s: shape of weight
output_shape: shape after deconv2d
'''
if output_shape == None:
x_shape = tf.shape(x)
output_shape = tf.stack(
[x_shape[0], x_shape[1] * 2, x_shape[2] * 2, x_shape[3] // 2])
W_t = utils.weight_variable(W_s, stddev=stddev, name='W_' + name)
b_t = utils.bias_variable([W_s[2]], name='b_' + name)
#conv_t = utils.conv2d_transpose_strided_valid(x, W_t, b_t, output_shape, stride)
conv_t = utils.conv2d_transpose_strided(x, W_t, b_t, output_shape, stride)
if if_relu:
conv_t = tf.nn.relu(conv_t, name=name + '_relu')
conv_concat = utils.crop_and_concat(concat_x, conv_t)
return conv_concat
def inference(image, keep_prob):
"""
Semantic segmentation network definition
:param image: input image. Should have values in range 0-255
:param keep_prob:
:return:
"""
print("setting up vgg initialized conv layers ...")
model_data = utils.get_model_data(FLAGS.model_dir, FLAGS.MODEL_NAME)
#mean = model_data['normalization'][0][0][0]
#mean_pixel = np.mean(mean, axis=(0, 1))
weights = np.squeeze(model_data['params'][0])
#processed_image = utils.process_image(image, mean_pixel)
with tf.variable_scope("inference"):
image_net = res_net(weights, image)
#conv_final_layer = image_net["res5c"]
conv_final_layer = image_net
fc_w = utils.weight_variable([1, 1, 2048, FLAGS.NUM_OF_CLASSESS],
name="fc_w")
fc_b = utils.bias_variable([FLAGS.NUM_OF_CLASSESS], name="fc_b")
fc = utils.conv2d_basic(conv_final_layer, fc_w, fc_b)
fc_dropout = tf.nn.dropout(fc, keep_prob)
logits = tf.squeeze(fc_dropout, [1, 2])
#logits = tf.squeeze(utils.conv2d_basic(conv_final_layer, fc_w, fc_b), [1,2])
print('logits shape', logits.shape)
annotation_pred = tf.argmax(logits, dimension=1, name="prediction")
return tf.expand_dims(annotation_pred, dim=1), logits
def conv2d_layer(x,
name,
W_s,
pool_,
if_relu=False,
stride=2,
stddev=0.02,
if_dropout=False,
keep_prob_=1):
'''Conv2d operator
Args:
pool_: if pool_==0:not pooling else pooling
'''
W = utils.weight_variable(W_s, stddev=stddev, name='W' + name)
b = utils.bias_variable([W_s[3]], name='b' + name)
#conv = utils.conv2d_strided_valid(x, W, b, stride)
conv = utils.conv2d_strided(x, W, b, stride)
print('shape after conv: ', conv.shape)
print('--------------------------------')
if if_dropout:
conv = tf.nn.dropout(conv, keep_prob_)
if if_relu:
conv = tf.nn.relu(conv, name=name + '_relu')
if pool_:
conv = utils.max_pool(conv, pool_, 2)
print('shape after pool: ', conv.shape)
return conv
def conv_layer(input, r_field, input_c, out_c, nr):
W = utils.weight_variable([r_field, r_field, input_c, out_c],
name="W" + str(nr))
b = utils.bias_variable([out_c], name="b" + str(nr))
conv = utils.conv2d_basic(input, W, b, name="conv" + str(nr))
relu = tf.nn.relu(conv, name="relu" + str(nr))
return relu
def build_pose_layers(self, image, keep_prob, rois):
TRUNCATE = 0
with tf.variable_scope("pose"):
shape6 = tf.shape(self.image_net["conv5_3"])
deconv_shape6 = tf.stack([shape6[0], 56, 56, 512])
W_6 = utils.weight_variable([4, 4, 512, 512], name="W_6")
b_6 = utils.bias_variable([512], name="b_6")
conv_6 = utils.conv2d_transpose_strided(self.image_net["conv5_3"], W_6, b_6, output_shape=deconv_shape6, stride=4)
roi_layer6 = ROIPoolingLayer(self.roi_pool_h, self.roi_pool_w)
pooled_features6 = roi_layer6([conv_6, rois])
pooled_features6 = tf.nn.dropout(pooled_features6, keep_prob=keep_prob)
shape7 = tf.shape(self.image_net["conv4_3"])
deconv_shape7 = tf.stack([shape7[0], 56, 56, 512])
W_7 = utils.weight_variable([2, 2, 512, 512], name="W_7")
b_7 = utils.bias_variable([512], name="b_7")
conv_7 = utils.conv2d_transpose_strided(self.image_net["conv4_3"], W_7, b_7, output_shape=deconv_shape7, stride=2)
roi_layer7 = ROIPoolingLayer(self.roi_pool_h, self.roi_pool_w)
pooled_features7 = roi_layer7([conv_7, rois])
pooled_features7 = tf.nn.dropout(pooled_features7, keep_prob=keep_prob)
roi_add8 = tf.keras.layers.Add()([pooled_features6, pooled_features7])
roi_add9 = tf.reduce_sum(roi_add8, axis=1)
shape = roi_add9.get_shape().as_list()
dim = 1
for d in shape[1:]:
dim *= d
roi_add9 = tf.reshape(roi_add9, [-1, dim])
fc9_w = tf.reshape(self.vgg_fc["fc6"][0], [dim, 4096])
fc9 = tf.nn.bias_add(tf.matmul(roi_add9, fc9_w), self.vgg_fc["fc6"][1])
fc_dropout9 = tf.nn.dropout(fc9, keep_prob=keep_prob)
fc10_w = tf.reshape(self.vgg_fc["fc7"][0], [4096, 4096])
fc10 = tf.nn.bias_add(tf.matmul(fc_dropout9, fc10_w), self.vgg_fc["fc7"][1])
fc_dropout10 = tf.nn.dropout(fc10, keep_prob=keep_prob)
W11 = utils.weight_variable([4096, 4 * (self.n_classes - 1)], name="W11")
b11 = utils.bias_variable([4 * (self.n_classes - 1)], name="b11")
fc11 = tf.nn.bias_add(tf.matmul(fc_dropout10, W11), b11)
tanh11 = tf.math.tanh(fc11)
return tanh11
def inference_fully_convolutional(dataset):
'''
Fully convolutional inference on notMNIST dataset
:param datset: [batch_size, 28*28*1] tensor
:return: logits
'''
dataset_reshaped = tf.reshape(dataset, [-1, 28, 28, 1])
with tf.name_scope("conv1") as scope:
W_conv1 = utils.weight_variable_xavier_initialized([3, 3, 1, 32],
name="W_conv1")
b_conv1 = utils.bias_variable([32], name="b_conv1")
h_conv1 = tf.nn.relu(
utils.conv2d_strided(dataset_reshaped, W_conv1, b_conv1))
with tf.name_scope("conv2") as scope:
W_conv2 = utils.weight_variable_xavier_initialized([3, 3, 32, 64],
name="W_conv2")
b_conv2 = utils.bias_variable([64], name="b_conv2")
h_conv2 = tf.nn.relu(utils.conv2d_strided(h_conv1, W_conv2, b_conv2))
with tf.name_scope("conv3") as scope:
W_conv3 = utils.weight_variable_xavier_initialized([3, 3, 64, 128],
name="W_conv3")
b_conv3 = utils.bias_variable([128], name="b_conv3")
h_conv3 = tf.nn.relu(utils.conv2d_strided(h_conv2, W_conv3, b_conv3))
with tf.name_scope("conv4") as scope:
W_conv4 = utils.weight_variable_xavier_initialized([3, 3, 128, 256],
name="W_conv4")
b_conv4 = utils.bias_variable([256], name="b_conv4")
h_conv4 = tf.nn.relu(utils.conv2d_strided(h_conv3, W_conv4, b_conv4))
with tf.name_scope("conv5") as scope:
# W_conv5 = utils.weight_variable_xavier_initialized([2, 2, 256, 512], name="W_conv5")
# b_conv5 = utils.bias_variable([512], name="b_conv5")
# h_conv5 = tf.nn.relu(utils.conv2d_strided(h_conv4, W_conv5, b_conv5))
h_conv5 = utils.avg_pool_2x2(h_conv4)
with tf.name_scope("conv6") as scope:
W_conv6 = utils.weight_variable_xavier_initialized([1, 1, 256, 10],
name="W_conv6")
b_conv6 = utils.bias_variable([10], name="b_conv6")
logits = tf.nn.relu(utils.conv2d_basic(h_conv5, W_conv6, b_conv6))
print logits.get_shape()
logits = tf.reshape(logits, [-1, 10])
return logits
def decoder_fc(z):
with tf.variable_scope("decoder") as scope:
Wd_fc1 = utils.weight_variable([FLAGS.z_dim, 50], name="Wd_fc1")
bd_fc1 = utils.bias_variable([50], name="bd_fc1")
hd_relu1 = activation_function(tf.matmul(z, Wd_fc1) + bd_fc1,
name="hdfc_1")
Wd_fc2 = utils.weight_variable([50, 50], name="Wd_fc2")
bd_fc2 = utils.bias_variable([50], name="bd_fc2")
hd_relu2 = activation_function(tf.matmul(hd_relu1, Wd_fc2) + bd_fc2,
name="hdfc_2")
Wd_fc3 = utils.weight_variable([50, IMAGE_SIZE * IMAGE_SIZE],
name="Wd_fc3")
bd_fc3 = utils.bias_variable([IMAGE_SIZE * IMAGE_SIZE], name="bd_fc3")
pred_image = tf.matmul(hd_relu2, Wd_fc3) + bd_fc3
return pred_image
def inference(image, keep_prob):
"""
Semantic segmentation network definition
:param image: input image. Should have values in range 0-255
:param keep_prob:
:return:
"""
print("setting up vgg initialized conv layers ...")
model_data = utils.get_model_data(FLAGS.model_dir, MODEL_URL)
mean = model_data['normalization'][0][0][0]
mean_pixel = np.mean(mean, axis=(0, 1))
weights = np.squeeze(model_data['layers'])
processed_image = utils.process_image(image, mean_pixel)
with tf.variable_scope("inference"):
vgg_end_layer = 'conv4_4'
image_net = vgg_net(weights, processed_image, end_layer=vgg_end_layer)
conv_final_layer = image_net[vgg_end_layer]
dropout = tf.nn.dropout(conv_final_layer, keep_prob=keep_prob)
W_final = utils.weight_variable([1, 1, 512, NUM_OF_CLASSES], name="W_final")
b_final = utils.bias_variable([NUM_OF_CLASSES], name="b_final")
conv_final = utils.conv2d_basic(dropout, W_final, b_final)
if FLAGS.debug:
utils.add_activation_summary(conv_final)
# now to upscale to actual image size
deconv_shape2 = image_net["pool2"].get_shape()
W_t2 = utils.weight_variable([4, 4, deconv_shape2[3].value, NUM_OF_CLASSES], name="W_t2")
b_t2 = utils.bias_variable([deconv_shape2[3].value], name="b_t2")
conv_t2 = utils.conv2d_transpose_strided(conv_final, W_t2, b_t2, output_shape=tf.shape(image_net["pool2"]))
fuse_2 = tf.add(conv_t2, image_net["pool2"], name="fuse_2")
shape = tf.shape(image)
deconv_shape3 = tf.stack([shape[0], shape[1], shape[2], NUM_OF_CLASSES])
W_t3 = utils.weight_variable([8, 8, NUM_OF_CLASSES, deconv_shape2[3].value], name="W_t3")
b_t3 = utils.bias_variable([NUM_OF_CLASSES], name="b_t3")
conv_t3 = utils.conv2d_transpose_strided(fuse_2, W_t3, b_t3, output_shape=deconv_shape3, stride=4)
annotation_pred = tf.argmax(conv_t3, axis=3, name="prediction", output_type=tf.int32)
return tf.expand_dims(annotation_pred, axis=3), conv_t3
def vgg_net(weights, image):
layers = ('conv1_1', 'relu1_1', 'conv1_2', 'relu1_2', 'pool1', 'conv2_1',
'relu2_1', 'conv2_2', 'relu2_2', 'pool2', 'conv3_1', 'relu3_1',
'conv3_2', 'relu3_2', 'conv3_3', 'relu3_3', 'conv3_4', 'relu3_4',
'pool3', 'conv4_1', 'relu4_1', 'conv4_2', 'relu4_2', 'conv4_3',
'relu4_3', 'conv4_4', 'relu4_4', 'pool4', 'conv5_1', 'relu5_1',
'conv5_2', 'relu5_2', 'conv5_3', 'relu5_3', 'conv5_4', 'relu5_4')
'''
weights[i][0][0][0][0]:
<tf.Variable 'inference/conv1_1_w:0' shape=(3, 3, 3, 64) dtype=float32_ref>
<tf.Variable 'inference/conv1_1_b:0' shape=(64,) dtype=float32_ref>
<tf.Variable 'inference/conv1_2_w:0' shape=(3, 3, 64, 64) dtype=float32_ref>
<tf.Variable 'inference/conv1_2_b:0' shape=(64,) dtype=float32_ref>
<tf.Variable 'inference/conv2_1_w:0' shape=(3, 3, 64, 128) dtype=float32_ref>
<tf.Variable 'inference/conv2_1_b:0' shape=(128,) dtype=float32_ref>
<tf.Variable 'inference/conv2_2_w:0' shape=(3, 3, 128, 128) dtype=float32_ref>
<tf.Variable 'inference/conv2_2_b:0' shape=(128,) dtype=float32_ref>
'''
net = {}
current = image
for i, name in enumerate(
layers
): # 对于一个可迭代/可遍历的对象（如列表、字符串），enumerate将其组成一个索引序列，利用它可以同时获得索引和值
kind = name[:4]
num = name[4:]
if kind == 'conv' and num == '1_1':
W = utils.weight_variable(
[3, 3, 4, 64],
name=name + "_w") # [patch 7*7,insize 512, outsize 4096]
b = utils.bias_variable([64], name=name + "_b")
current = utils.conv2d_basic(current, W, b)
elif kind == 'conv' and num != '1_1':
kernels, bias = weights[i][0][0][0][0]
# print("kernels:",i,kernels)
# print kernels
# matconvnet: weights are [width, height, in_channels, out_channels]
# tensorflow: weights are [height, width, in_channels, out_channels]
kernels = utils.get_variable(np.transpose(kernels, (1, 0, 2, 3)),
name=name + "_w")
# print(kernels)
bias = utils.get_variable(bias.reshape(-1), name=name + "_b")
# print(bias)
current = utils.conv2d_basic(current, kernels, bias)
elif kind == 'relu':
current = tf.nn.relu(current, name=name)
if FLAGS.debug:
utils.add_activation_summary(current)
elif kind == 'pool':
current = utils.avg_pool_2x2(current)
net[name] = current
return net
def deepscores_cnn(image, nr_class):
# placeholder for dropout input
keep_prob = tf.placeholder(tf.float32)
# five layers of 3x3 convolutions, followed by relu, 2x2-maxpool and dropout
W1 = utils.weight_variable([3, 3, 1, 32], name="W1")
b1 = utils.bias_variable([32], name="b1")
conv1 = utils.conv2d_basic(image, W1, b1, name="conv1")
relu1 = tf.nn.relu(conv1, name="relu1")
pool1 = utils.max_pool_2x2(relu1)
dropout1 = tf.nn.dropout(pool1, keep_prob=keep_prob)
W2 = utils.weight_variable([3, 3, 32, 64], name="W2")
b2 = utils.bias_variable([64], name="b2")
conv2 = utils.conv2d_basic(dropout1, W2, b2, name="conv2")
relu2 = tf.nn.relu(conv2, name="relu2")
pool2 = utils.max_pool_2x2(relu2)
dropout2 = tf.nn.dropout(pool2, keep_prob=keep_prob)
W3 = utils.weight_variable([3, 3, 64, 128], name="W3")
b3 = utils.bias_variable([128], name="b3")
conv3 = utils.conv2d_basic(dropout2, W3, b3, name="conv3")
relu3 = tf.nn.relu(conv3, name="relu3")
pool3 = utils.max_pool_2x2(relu3)
dropout3 = tf.nn.dropout(pool3, keep_prob=keep_prob)
W4 = utils.weight_variable([3, 3, 128, 256], name="W4")
b4 = utils.bias_variable([256], name="b4")
conv4 = utils.conv2d_basic(dropout3, W4, b4, name="conv4")
relu4 = tf.nn.relu(conv4, name="relu4")
pool4 = utils.max_pool_2x2(relu4)
dropout4 = tf.nn.dropout(pool4, keep_prob=keep_prob)
W5 = utils.weight_variable([3, 3, 256, 512], name="W5")
b5 = utils.bias_variable([512], name="b5")
conv5 = utils.conv2d_basic(dropout4, W5, b5, name="conv5")
relu5 = tf.nn.relu(conv5, name="relu5")
pool5 = utils.max_pool_2x2(relu5)
dropout5 = tf.nn.dropout(pool5, keep_prob=keep_prob)
# to fully connected layers
# downsampled 5 times so feature maps should be 32 times smaller
# size is 7*4*512
W_fc1 = utils.weight_variable([7*4*512, 1024])
b_fc1 = utils.bias_variable([1024])
dropout5_flat = tf.reshape(dropout5, [-1, 7*4*512])
h_fc1 = tf.nn.relu(tf.matmul(dropout5_flat, W_fc1) + b_fc1)
W_fc2 = utils.weight_variable([1024, nr_class])
b_fc2 = utils.bias_variable([nr_class])
y_conv = tf.matmul(h_fc1, W_fc2) + b_fc2
return y_conv, keep_prob
def discriminator(input_images, train_mode):
# dropout_prob = 1.0
# if train_mode:
# dropout_prob = 0.5
W_conv0 = utils.weight_variable([5, 5, NUM_OF_CHANNELS, 64 * 1],
name="W_conv0")
b_conv0 = utils.bias_variable([64 * 1], name="b_conv0")
h_conv0 = utils.conv2d_strided(input_images, W_conv0, b_conv0)
h_bn0 = h_conv0 # utils.batch_norm(h_conv0, 64 * 1, train_mode, scope="disc_bn0")
h_relu0 = utils.leaky_relu(h_bn0, 0.2, name="h_relu0")
utils.add_activation_summary(h_relu0)
W_conv1 = utils.weight_variable([5, 5, 64 * 1, 64 * 2], name="W_conv1")
b_conv1 = utils.bias_variable([64 * 2], name="b_conv1")
h_conv1 = utils.conv2d_strided(h_relu0, W_conv1, b_conv1)
h_bn1 = utils.batch_norm(h_conv1, 64 * 2, train_mode, scope="disc_bn1")
h_relu1 = utils.leaky_relu(h_bn1, 0.2, name="h_relu1")
utils.add_activation_summary(h_relu1)
W_conv2 = utils.weight_variable([5, 5, 64 * 2, 64 * 4], name="W_conv2")
b_conv2 = utils.bias_variable([64 * 4], name="b_conv2")
h_conv2 = utils.conv2d_strided(h_relu1, W_conv2, b_conv2)
h_bn2 = utils.batch_norm(h_conv2, 64 * 4, train_mode, scope="disc_bn2")
h_relu2 = utils.leaky_relu(h_bn2, 0.2, name="h_relu2")
utils.add_activation_summary(h_relu2)
W_conv3 = utils.weight_variable([5, 5, 64 * 4, 64 * 8], name="W_conv3")
b_conv3 = utils.bias_variable([64 * 8], name="b_conv3")
h_conv3 = utils.conv2d_strided(h_relu2, W_conv3, b_conv3)
h_bn3 = utils.batch_norm(h_conv3, 64 * 8, train_mode, scope="disc_bn3")
h_relu3 = utils.leaky_relu(h_bn3, 0.2, name="h_relu3")
utils.add_activation_summary(h_relu3)
shape = h_relu3.get_shape().as_list()
h_3 = tf.reshape(
h_relu3,
[FLAGS.batch_size, (IMAGE_SIZE // 16) * (IMAGE_SIZE // 16) * shape[3]])
W_4 = utils.weight_variable([h_3.get_shape().as_list()[1], 1], name="W_4")
b_4 = utils.bias_variable([1], name="b_4")
h_4 = tf.matmul(h_3, W_4) + b_4
return tf.nn.sigmoid(h_4), h_4, h_relu3
def encoder_fc(images):
with tf.variable_scope("encoder") as scope:
W_fc1 = utils.weight_variable([IMAGE_SIZE * IMAGE_SIZE, 50], name="W_fc1")
b_fc1 = utils.bias_variable([50], name="b_fc1")
h_relu1 = activation_function(tf.matmul(images, W_fc1) + b_fc1, name="hfc_1")
W_fc2 = utils.weight_variable([50, 50], name="W_fc2")
b_fc2 = utils.bias_variable([50], name="b_fc2")
h_relu2 = activation_function(tf.matmul(h_relu1, W_fc2) + b_fc2, name="hfc_2")
W_fc3 = utils.weight_variable([50, FLAGS.z_dim], name="W_fc3")
b_fc3 = utils.bias_variable([FLAGS.z_dim], name="b_fc3")
mu = tf.add(tf.matmul(h_relu2, W_fc3), b_fc3, name="mu")
utils.add_activation_summary(mu)
W_fc4 = utils.weight_variable([50, FLAGS.z_dim], name="W_fc4")
b_fc4 = utils.bias_variable([FLAGS.z_dim], name="b_fc4")
log_var = tf.add(tf.matmul(h_relu2, W_fc4), b_fc4, name="log_var")
utils.add_activation_summary(log_var)
return mu, log_var
def deconv_layer(input,
r_field,
in_channels,
out_channels,
out_shape,
nr,
stride=2):
W = utils.weight_variable([r_field, r_field, out_channels, in_channels],
name="W_t" + nr)
b = utils.bias_variable([out_channels], name="b_t" + nr)
conv_t1 = utils.conv2d_transpose_strided(input, W, b, out_shape)
return conv_t1
def inference_fully_convolutional(dataset):
'''
Fully convolutional inference on notMNIST dataset
:param datset: [batch_size, 28*28*1] tensor
:return: logits
'''
dataset_reshaped = tf.reshape(dataset, [-1, 28, 28, 1])
with tf.name_scope("conv1") as scope:
W_conv1 = utils.weight_variable_xavier_initialized([3, 3, 1, 32], name="W_conv1")
b_conv1 = utils.bias_variable([32], name="b_conv1")
h_conv1 = tf.nn.relu(utils.conv2d_strided(dataset_reshaped, W_conv1, b_conv1))
with tf.name_scope("conv2") as scope:
W_conv2 = utils.weight_variable_xavier_initialized([3, 3, 32, 64], name="W_conv2")
b_conv2 = utils.bias_variable([64], name="b_conv2")
h_conv2 = tf.nn.relu(utils.conv2d_strided(h_conv1, W_conv2, b_conv2))
with tf.name_scope("conv3") as scope:
W_conv3 = utils.weight_variable_xavier_initialized([3, 3, 64, 128], name="W_conv3")
b_conv3 = utils.bias_variable([128], name="b_conv3")
h_conv3 = tf.nn.relu(utils.conv2d_strided(h_conv2, W_conv3, b_conv3))
with tf.name_scope("conv4") as scope:
W_conv4 = utils.weight_variable_xavier_initialized([3, 3, 128, 256], name="W_conv4")
b_conv4 = utils.bias_variable([256], name="b_conv4")
h_conv4 = tf.nn.relu(utils.conv2d_strided(h_conv3, W_conv4, b_conv4))
with tf.name_scope("conv5") as scope:
# W_conv5 = utils.weight_variable_xavier_initialized([2, 2, 256, 512], name="W_conv5")
# b_conv5 = utils.bias_variable([512], name="b_conv5")
# h_conv5 = tf.nn.relu(utils.conv2d_strided(h_conv4, W_conv5, b_conv5))
h_conv5 = utils.avg_pool_2x2(h_conv4)
with tf.name_scope("conv6") as scope:
W_conv6 = utils.weight_variable_xavier_initialized([1, 1, 256, 10], name="W_conv6")
b_conv6 = utils.bias_variable([10], name="b_conv6")
logits = tf.nn.relu(utils.conv2d_basic(h_conv5, W_conv6, b_conv6))
print logits.get_shape()
logits = tf.reshape(logits, [-1, 10])
return logits
def down_layer(x, in_size, out_size, i):
# Down 1
W1 = utils.weight_variable([3, 3, 3, in_size, out_size // 2],
name="W_d_" + str(i) + "_1")
b1 = utils.bias_variable([out_size // 2],
name="b_d_" + str(i) + "_1")
conv1 = utils.conv3d_basic(x, W1, b1)
relu1 = tf.nn.relu(conv1, name="relu_d_" + str(i) + "_1")
relu1 = tf.nn.dropout(relu1, keep_prob=keep_prob)
# Down 2
W2 = utils.weight_variable([3, 3, 3, out_size // 2, out_size],
name="W_d_" + str(i) + "_2")
b2 = utils.bias_variable([out_size], name="b_d_" + str(i) + "_2")
conv2 = utils.conv3d_basic(relu1, W2, b2)
relu2 = tf.nn.relu(conv2, name="relu_d_" + str(i) + "_2")
relu2 = tf.nn.dropout(relu2, keep_prob=keep_prob)
# Pool 1
pool = utils.max_pool_2x2x2(relu2)
return relu2, pool
def discriminator(input_images, train_mode):
# dropout_prob = 1.0
# if train_mode:
# dropout_prob = 0.5
W_conv0 = utils.weight_variable([5, 5, NUM_OF_CHANNELS, 64 * 1], name="W_conv0")
b_conv0 = utils.bias_variable([64 * 1], name="b_conv0")
h_conv0 = utils.conv2d_strided(input_images, W_conv0, b_conv0)
h_bn0 = h_conv0 # utils.batch_norm(h_conv0, 64 * 1, train_mode, scope="disc_bn0")
h_relu0 = utils.leaky_relu(h_bn0, 0.2, name="h_relu0")
utils.add_activation_summary(h_relu0)
W_conv1 = utils.weight_variable([5, 5, 64 * 1, 64 * 2], name="W_conv1")
b_conv1 = utils.bias_variable([64 * 2], name="b_conv1")
h_conv1 = utils.conv2d_strided(h_relu0, W_conv1, b_conv1)
h_bn1 = utils.batch_norm(h_conv1, 64 * 2, train_mode, scope="disc_bn1")
h_relu1 = utils.leaky_relu(h_bn1, 0.2, name="h_relu1")
utils.add_activation_summary(h_relu1)
W_conv2 = utils.weight_variable([5, 5, 64 * 2, 64 * 4], name="W_conv2")
b_conv2 = utils.bias_variable([64 * 4], name="b_conv2")
h_conv2 = utils.conv2d_strided(h_relu1, W_conv2, b_conv2)
h_bn2 = utils.batch_norm(h_conv2, 64 * 4, train_mode, scope="disc_bn2")
h_relu2 = utils.leaky_relu(h_bn2, 0.2, name="h_relu2")
utils.add_activation_summary(h_relu2)
W_conv3 = utils.weight_variable([5, 5, 64 * 4, 64 * 8], name="W_conv3")
b_conv3 = utils.bias_variable([64 * 8], name="b_conv3")
h_conv3 = utils.conv2d_strided(h_relu2, W_conv3, b_conv3)
h_bn3 = utils.batch_norm(h_conv3, 64 * 8, train_mode, scope="disc_bn3")
h_relu3 = utils.leaky_relu(h_bn3, 0.2, name="h_relu3")
utils.add_activation_summary(h_relu3)
shape = h_relu3.get_shape().as_list()
h_3 = tf.reshape(h_relu3, [FLAGS.batch_size, (IMAGE_SIZE // 16) * (IMAGE_SIZE // 16) * shape[3]])
W_4 = utils.weight_variable([h_3.get_shape().as_list()[1], 1], name="W_4")
b_4 = utils.bias_variable([1], name="b_4")
h_4 = tf.matmul(h_3, W_4) + b_4
return tf.nn.sigmoid(h_4), h_4, h_relu3
def inpainter(embedding, train_mode):
with tf.variable_scope("context_inpainter"):
image_size = IMAGE_SIZE // 32
with tf.name_scope("dec_fc") as scope:
W_fc = utils.weight_variable([1024, image_size * image_size * 512], name="W_fc")
b_fc = utils.bias_variable([image_size * image_size * 512], name="b_fc")
h_fc = tf.nn.relu(tf.matmul(embedding, W_fc) + b_fc)
with tf.name_scope("dec_conv1") as scope:
h_reshaped = tf.reshape(h_fc, tf.pack([tf.shape(h_fc)[0], image_size, image_size, 512]))
W_conv_t1 = utils.weight_variable_xavier_initialized([3, 3, 256, 512], name="W_conv_t1")
b_conv_t1 = utils.bias_variable([256], name="b_conv_t1")
deconv_shape = tf.pack([tf.shape(h_reshaped)[0], 2 * image_size, 2 * image_size, 256])
h_conv_t1 = utils.conv2d_transpose_strided(h_reshaped, W_conv_t1, b_conv_t1, output_shape=deconv_shape)
h_bn_t1 = utils.batch_norm(h_conv_t1, 256, train_mode, scope="conv_t1_bn")
h_relu_t1 = tf.nn.relu(h_bn_t1)
with tf.name_scope("dec_conv2") as scope:
W_conv_t2 = utils.weight_variable_xavier_initialized([3, 3, 128, 256], name="W_conv_t2")
b_conv_t2 = utils.bias_variable([128], name="b_conv_t2")
deconv_shape = tf.pack([tf.shape(h_relu_t1)[0], 4 * image_size, 4 * image_size, 128])
h_conv_t2 = utils.conv2d_transpose_strided(h_relu_t1, W_conv_t2, b_conv_t2, output_shape=deconv_shape)
h_bn_t2 = utils.batch_norm(h_conv_t2, 128, train_mode, scope="conv_t2_bn")
h_relu_t2 = tf.nn.relu(h_bn_t2)
with tf.name_scope("dec_conv3") as scope:
W_conv_t3 = utils.weight_variable_xavier_initialized([3, 3, 64, 128], name="W_conv_t3")
b_conv_t3 = utils.bias_variable([64], name="b_conv_t3")
deconv_shape = tf.pack([tf.shape(h_relu_t2)[0], 8 * image_size, 8 * image_size, 64])
h_conv_t3 = utils.conv2d_transpose_strided(h_relu_t2, W_conv_t3, b_conv_t3, output_shape=deconv_shape)
h_bn_t3 = utils.batch_norm(h_conv_t3, 64, train_mode, scope="conv_t3_bn")
h_relu_t3 = tf.nn.relu(h_bn_t3)
with tf.name_scope("dec_conv4") as scope:
W_conv_t4 = utils.weight_variable_xavier_initialized([3, 3, 3, 64], name="W_conv_t4")
b_conv_t4 = utils.bias_variable([3], name="b_conv_t4")
deconv_shape = tf.pack([tf.shape(h_relu_t3)[0], 16 * image_size, 16 * image_size, 3])
pred_image = utils.conv2d_transpose_strided(h_relu_t3, W_conv_t4, b_conv_t4, output_shape=deconv_shape)
return pred_image
def deconv2d_layer(x, name, W_s, output_shape=None, stride=2):
'''Deconv2d operator
Args:
x: inputs
W_s: shape of weight
output_shape: shape after deconv2d
'''
W_t = utils.weight_variable(W_s, name='W_' + name)
b_t = utils.bias_variable([W_s[2]], name='b_' + name)
conv_t = utils.conv2d_transpose_strided(x, W_t, b_t, output_shape, stride)
print('conv_%s: ' % name, conv_t.get_shape())
return conv_t
def inferece(dataset, prob):
with tf.name_scope("conv1") as scope:
W_conv1 = utils.weight_variable([5, 5, 1, 32])
b_conv1 = utils.bias_variable([32])
tf.histogram_summary("W_conv1", W_conv1)
tf.histogram_summary("b_conv1", b_conv1)
h_conv1 = utils.conv2d_basic(dataset, W_conv1, b_conv1)
h_1 = tf.nn.relu(h_conv1)
h_pool1 = utils.max_pool_2x2(h_1)
add_to_regularization_loss(W_conv1, b_conv1)
with tf.name_scope("conv2") as scope:
W_conv2 = utils.weight_variable([3, 3, 32, 64])
b_conv2 = utils.bias_variable([64])
tf.histogram_summary("W_conv2", W_conv2)
tf.histogram_summary("b_conv2", b_conv2)
h_conv2 = utils.conv2d_basic(h_pool1, W_conv2, b_conv2)
h_2 = tf.nn.relu(h_conv2)
h_pool2 = utils.max_pool_2x2(h_2)
add_to_regularization_loss(W_conv2, b_conv2)
with tf.name_scope("fc_1") as scope:
image_size = IMAGE_SIZE / 4
h_flat = tf.reshape(h_pool2, [-1, image_size * image_size * 64])
W_fc1 = utils.weight_variable([image_size * image_size * 64, 256])
b_fc1 = utils.bias_variable([256])
tf.histogram_summary("W_fc1", W_fc1)
tf.histogram_summary("b_fc1", b_fc1)
h_fc1 = tf.nn.relu(tf.matmul(h_flat, W_fc1) + b_fc1)
h_fc1_dropout = tf.nn.dropout(h_fc1, prob)
with tf.name_scope("fc_2") as scope:
W_fc2 = utils.weight_variable([256, NUM_LABELS])
b_fc2 = utils.bias_variable([NUM_LABELS])
tf.histogram_summary("W_fc2", W_fc2)
tf.histogram_summary("b_fc2", b_fc2)
pred = tf.matmul(h_fc1, W_fc2) + b_fc2
return pred
def inference(dataset):
with tf.name_scope("conv1") as scope:
W1 = utils.weight_variable([5, 5, 1, 32], name="W1")
b1 = utils.bias_variable([32], name="b1")
tf.histogram_summary("W1", W1)
tf.histogram_summary("b1", b1)
h_conv1 = utils.conv2d_basic(dataset, W1, b1)
h_norm1 = utils.local_response_norm(h_conv1)
h_1 = tf.nn.relu(h_norm1, name="conv1")
h_pool1 = utils.max_pool_2x2(h_1)
with tf.name_scope("conv2") as scope:
W2 = utils.weight_variable([3, 3, 32, 64], name="W2")
b2 = utils.bias_variable([64], name="b2")
tf.histogram_summary("W2", W2)
tf.histogram_summary("b2", b2)
h_conv2 = utils.conv2d_basic(h_pool1, W2, b2)
h_norm2 = utils.local_response_norm(h_conv2)
h_2 = tf.nn.relu(h_norm2, name="conv2")
h_pool2 = utils.max_pool_2x2(h_2)
with tf.name_scope("conv3") as scope:
W3 = utils.weight_variable([3, 3, 64, 128], name="W3")
b3 = utils.bias_variable([128], name="b3")
tf.histogram_summary("W3", W3)
tf.histogram_summary("b3", b3)
h_conv3 = utils.conv2d_basic(h_pool2, W3, b3)
h_norm3 = utils.local_response_norm(h_conv3)
h_3 = tf.nn.relu(h_norm3, name="conv3")
h_pool3 = utils.max_pool_2x2(h_3)
with tf.name_scope("conv4") as scope:
W4 = utils.weight_variable([3, 3, 128, 256], name="W4")
b4 = utils.bias_variable([256], name="b4")
tf.histogram_summary("W4", W4)
tf.histogram_summary("b4", b4)
h_conv4 = utils.conv2d_basic(h_pool3, W4, b4)
h_norm4 = utils.local_response_norm(h_conv4)
h_4 = tf.nn.relu(h_norm4, name="conv4")
with tf.name_scope("fc1") as scope:
image_size = IMAGE_SIZE // 8
h_flat = tf.reshape(h_4, [-1, image_size * image_size * 256])
W_fc1 = utils.weight_variable([image_size * image_size * 256, 512], name="W_fc1")
b_fc1 = utils.bias_variable([512], name="b_fc1")
tf.histogram_summary("W_fc1", W_fc1)
tf.histogram_summary("b_fc1", b_fc1)
h_fc1 = tf.nn.relu(tf.matmul(h_flat, W_fc1) + b_fc1)
with tf.name_scope("fc2") as scope:
W_fc2 = utils.weight_variable([512, NUM_LABELS], name="W_fc2")
b_fc2 = utils.bias_variable([NUM_LABELS], name="b_fc2")
tf.histogram_summary("W_fc2", W_fc2)
tf.histogram_summary("b_fc2", b_fc2)
pred = tf.matmul(h_fc1, W_fc2) + b_fc2
return pred
def decoder_conv(embedding):
image_size = IMAGE_SIZE // 16
with tf.name_scope("dec_fc") as scope:
W_fc1 = utils.weight_variable([512, image_size * image_size * 256], name="W_fc1")
b_fc1 = utils.bias_variable([image_size * image_size * 256], name="b_fc1")
h_fc1 = tf.nn.relu(tf.matmul(embedding, W_fc1) + b_fc1)
with tf.name_scope("dec_conv1") as scope:
h_reshaped = tf.reshape(h_fc1, tf.pack([tf.shape(h_fc1)[0], image_size, image_size, 256]))
W_conv_t1 = utils.weight_variable([3, 3, 128, 256], name="W_conv_t1")
b_conv_t1 = utils.bias_variable([128], name="b_conv_t1")
deconv_shape = tf.pack([tf.shape(h_fc1)[0], 2 * image_size, 2 * image_size, 128])
h_conv_t1 = tf.nn.relu(
utils.conv2d_transpose_strided(h_reshaped, W_conv_t1, b_conv_t1, output_shape=deconv_shape))
with tf.name_scope("dec_conv2") as scope:
W_conv_t2 = utils.weight_variable([3, 3, 64, 128], name="W_conv_t2")
b_conv_t2 = utils.bias_variable([64], name="b_conv_t2")
deconv_shape = tf.pack([tf.shape(h_conv_t1)[0], 4 * image_size, 4 * image_size, 64])
h_conv_t2 = tf.nn.relu(
utils.conv2d_transpose_strided(h_conv_t1, W_conv_t2, b_conv_t2, output_shape=deconv_shape))
with tf.name_scope("dec_conv3") as scope:
W_conv_t3 = utils.weight_variable([3, 3, 32, 64], name="W_conv_t3")
b_conv_t3 = utils.bias_variable([32], name="b_conv_t3")
deconv_shape = tf.pack([tf.shape(h_conv_t2)[0], 8 * image_size, 8 * image_size, 32])
h_conv_t3 = tf.nn.relu(
utils.conv2d_transpose_strided(h_conv_t2, W_conv_t3, b_conv_t3, output_shape=deconv_shape))
with tf.name_scope("dec_conv4") as scope:
W_conv_t4 = utils.weight_variable([3, 3, 3, 32], name="W_conv_t4")
b_conv_t4 = utils.bias_variable([3], name="b_conv_t4")
deconv_shape = tf.pack([tf.shape(h_conv_t3)[0], IMAGE_SIZE, IMAGE_SIZE, 3])
pred_image = utils.conv2d_transpose_strided(h_conv_t3, W_conv_t4, b_conv_t4, output_shape=deconv_shape)
return pred_image
def encoder(dataset, train_mode):
with tf.variable_scope("Encoder"):
with tf.name_scope("enc_conv1") as scope:
W_conv1 = utils.weight_variable_xavier_initialized([3, 3, 3, 32], name="W_conv1")
b_conv1 = utils.bias_variable([32], name="b_conv1")
h_conv1 = utils.conv2d_strided(dataset, W_conv1, b_conv1)
h_bn1 = utils.batch_norm(h_conv1, 32, train_mode, scope="conv1_bn")
h_relu1 = tf.nn.relu(h_bn1)
with tf.name_scope("enc_conv2") as scope:
W_conv2 = utils.weight_variable_xavier_initialized([3, 3, 32, 64], name="W_conv2")
b_conv2 = utils.bias_variable([64], name="b_conv2")
h_conv2 = utils.conv2d_strided(h_relu1, W_conv2, b_conv2)
h_bn2 = utils.batch_norm(h_conv2, 64, train_mode, scope="conv2_bn")
h_relu2 = tf.nn.relu(h_bn2)
with tf.name_scope("enc_conv3") as scope:
W_conv3 = utils.weight_variable_xavier_initialized([3, 3, 64, 128], name="W_conv3")
b_conv3 = utils.bias_variable([128], name="b_conv3")
h_conv3 = utils.conv2d_strided(h_relu2, W_conv3, b_conv3)
h_bn3 = utils.batch_norm(h_conv3, 128, train_mode, scope="conv3_bn")
h_relu3 = tf.nn.relu(h_bn3)
with tf.name_scope("enc_conv4") as scope:
W_conv4 = utils.weight_variable_xavier_initialized([3, 3, 128, 256], name="W_conv4")
b_conv4 = utils.bias_variable([256], name="b_conv4")
h_conv4 = utils.conv2d_strided(h_relu3, W_conv4, b_conv4)
h_bn4 = utils.batch_norm(h_conv4, 256, train_mode, scope="conv4_bn")
h_relu4 = tf.nn.relu(h_bn4)
with tf.name_scope("enc_conv5") as scope:
W_conv5 = utils.weight_variable_xavier_initialized([3, 3, 256, 512], name="W_conv5")
b_conv5 = utils.bias_variable([512], name="b_conv5")
h_conv5 = utils.conv2d_strided(h_relu4, W_conv5, b_conv5)
h_bn5 = utils.batch_norm(h_conv5, 512, train_mode, scope="conv5_bn")
h_relu5 = tf.nn.relu(h_bn5)
with tf.name_scope("enc_fc") as scope:
image_size = IMAGE_SIZE // 32
h_relu5_flatten = tf.reshape(h_relu5, [-1, image_size * image_size * 512])
W_fc = utils.weight_variable([image_size * image_size * 512, 1024], name="W_fc")
b_fc = utils.bias_variable([1024], name="b_fc")
encoder_val = tf.matmul(h_relu5_flatten, W_fc) + b_fc
return encoder_val
def inference_strided(input_image):
W1 = utils.weight_variable([9, 9, 3, 32])
b1 = utils.bias_variable([32])
tf.histogram_summary("W1", W1)
tf.histogram_summary("b1", b1)
h_conv1 = tf.nn.relu(utils.conv2d_basic(input_image, W1, b1))
W2 = utils.weight_variable([3, 3, 32, 64])
b2 = utils.bias_variable([64])
tf.histogram_summary("W2", W2)
tf.histogram_summary("b2", b2)
h_conv2 = tf.nn.relu(utils.conv2d_strided(h_conv1, W2, b2))
W3 = utils.weight_variable([3, 3, 64, 128])
b3 = utils.bias_variable([128])
tf.histogram_summary("W3", W3)
tf.histogram_summary("b3", b3)
h_conv3 = tf.nn.relu(utils.conv2d_strided(h_conv2, W3, b3))
# upstrides
W4 = utils.weight_variable([3, 3, 64, 128])
b4 = utils.bias_variable([64])
tf.histogram_summary("W4", W4)
tf.histogram_summary("b4", b4)
# print h_conv3.get_shape()
# print W4.get_shape()
h_conv4 = tf.nn.relu(utils.conv2d_transpose_strided(h_conv3, W4, b4))
W5 = utils.weight_variable([3, 3, 32, 64])
b5 = utils.bias_variable([32])
tf.histogram_summary("W5", W5)
tf.histogram_summary("b5", b5)
h_conv5 = tf.nn.relu(utils.conv2d_transpose_strided(h_conv4, W5, b5))
W6 = utils.weight_variable([9, 9, 32, 3])
b6 = utils.bias_variable([3])
tf.histogram_summary("W6", W6)
tf.histogram_summary("b6", b6)
pred_image = tf.nn.tanh(utils.conv2d_basic(h_conv5, W6, b6))
return pred_image
def inference_fc(image):
with tf.name_scope("fc1") as scope:
W_fc1 = utils.weight_variable([IMAGE_SIZE * IMAGE_SIZE, 50], name="W_fc1")
b_fc1 = utils.bias_variable([50], name="b_fc1")
add_to_reg_loss_and_summary(W_fc1, b_fc1)
h_fc1 = tf.nn.tanh(tf.matmul(image, W_fc1) + b_fc1)
with tf.name_scope("fc2") as scope:
W_fc2 = utils.weight_variable([50, 50], name="W_fc2")
b_fc2 = utils.bias_variable([50], name="b_fc2")
add_to_reg_loss_and_summary(W_fc2, b_fc2)
h_fc2 = tf.nn.tanh(tf.matmul(h_fc1, W_fc2) + b_fc2)
with tf.name_scope("fc3") as scope:
W_fc3 = utils.weight_variable([50, 3], name="W_fc3")
b_fc3 = utils.bias_variable([3], name="b_fc3")
add_to_reg_loss_and_summary(W_fc3, b_fc3)
h_fc3 = tf.nn.tanh(tf.matmul(h_fc2, W_fc3) + b_fc3)
with tf.name_scope("fc4") as scope:
W_fc4 = utils.weight_variable([3, 50], name="W_fc4")
b_fc4 = utils.bias_variable([50], name="b_fc4")
add_to_reg_loss_and_summary(W_fc4, b_fc4)
h_fc4 = tf.nn.tanh(tf.matmul(h_fc3, W_fc4) + b_fc4)
with tf.name_scope("fc5") as scope:
W_fc5 = utils.weight_variable([50, 50], name="W_fc5")
b_fc5 = utils.bias_variable([50], name="b_fc5")
add_to_reg_loss_and_summary(W_fc5, b_fc5)
h_fc5 = tf.nn.tanh(tf.matmul(h_fc4, W_fc5) + b_fc5)
# h_fc_dropout = tf.nn.dropout(h_fc5, 0.5)
with tf.name_scope("fc6") as scope:
W_fc6 = utils.weight_variable([50, IMAGE_SIZE * IMAGE_SIZE], name="W_fc6")
b_fc6 = utils.bias_variable([IMAGE_SIZE * IMAGE_SIZE], name="b_fc6")
add_to_reg_loss_and_summary(W_fc6, b_fc6)
pred = tf.matmul(h_fc5, W_fc6) + b_fc6
return h_fc3, pred
def main(argv=None):
global_step = tf.Variable(0, trainable=False)
img_A = tf.placeholder(tf.float32, [None, IMAGE_SIZE, IMAGE_SIZE, 3])
img_B = tf.placeholder(tf.float32, [None, IMAGE_SIZE, IMAGE_SIZE, 3])
img_C = tf.placeholder(tf.float32, [None, IMAGE_SIZE, IMAGE_SIZE, 3])
img_D = tf.placeholder(tf.float32, [None, IMAGE_SIZE, IMAGE_SIZE, 3])
tf.image_summary("A", img_A, max_images=2)
tf.image_summary("B", img_B, max_images=2)
tf.image_summary("C", img_C, max_images=2)
tf.image_summary("Ground_truth", img_D, max_images=2)
print "Setting up encoder.."
with tf.variable_scope("encoder") as scope:
enc_A = encoder_conv(img_A)
scope.reuse_variables()
enc_B = encoder_conv(img_B)
enc_C = encoder_conv(img_C)
enc_D = encoder_conv(img_D)
print "Setting up analogy calc.."
# analogy calculation
analogy_input = tf.concat(1, [enc_B - enc_A, enc_C])
W_analogy1 = utils.weight_variable([1024, 512], name="W_analogy1")
b_analogy1 = utils.bias_variable([512], name="b_analogy1")
analogy_fc1 = tf.nn.relu(tf.matmul(analogy_input, W_analogy1) + b_analogy1)
W_analogy2 = utils.weight_variable([512, 512], name="W_analogy2")
b_analogy2 = utils.bias_variable([512], name="b_analogy2")
analogy_fc2 = tf.nn.relu(tf.matmul(analogy_fc1, W_analogy2) + b_analogy2)
pred = decoder_conv(enc_C + analogy_fc2)
tf.image_summary("Pred_image", pred, max_images=2)
print "Setting up regularization/ summary variables..."
for var in tf.trainable_variables():
add_to_regularization_and_summary(var)
print "Loss and train setup..."
loss1 = tf.sqrt(2*tf.nn.l2_loss(pred - img_D)) / FLAGS.batch_size
tf.scalar_summary("image_loss", loss1)
loss2 = tf.sqrt(2* tf.nn.l2_loss(enc_D - enc_C - analogy_fc2)) / FLAGS.batch_size
tf.scalar_summary("analogy_loss", loss2)
loss3 = tf.add_n(tf.get_collection("reg_loss"))
tf.scalar_summary("regularization", loss3)
total_loss = loss1 + ANALOGY_COEFF * loss2 + REGULARIZER * loss3
tf.scalar_summary("Total_loss", total_loss)
train_op = train(total_loss, global_step)
summary_op = tf.merge_all_summaries()
utils.maybe_download_and_extract(FLAGS.data_dir, DATA_URL, is_tarfile=True)
print "Initializing Loader class..."
loader = AnalogyDataLoader.Loader(FLAGS.data_dir, FLAGS.batch_size)
eval_A, eval_B, eval_C, eval_D = read_eval_inputs(loader)
eval_feed = {img_A: eval_A, img_B: eval_B, img_C: eval_C, img_D: eval_D}
with tf.Session() as sess:
sess.run(tf.initialize_all_variables())
print "Setting up summary and saver..."
summary_writer = tf.train.SummaryWriter(FLAGS.logs_dir, sess.graph)
saver = tf.train.Saver()
ckpt = tf.train.get_checkpoint_state(FLAGS.logs_dir)
if ckpt and ckpt.model_checkpoint_path:
saver.restore(sess, ckpt.model_checkpoint_path)
print "Model restored!"
for step in xrange(MAX_ITERATIONS):
A, B, C, D = read_train_inputs(loader)
feed_dict = {img_A: A, img_B: B, img_C: C, img_D: D}
if step % 1000 == 0:
eval_loss = sess.run([loss1, loss2, loss3, total_loss], feed_dict=eval_feed)
print "Evaluation: (Image loss %f, Variation loss %f, Reg loss %f) total loss %f" % tuple(eval_loss)
sess.run(train_op, feed_dict=feed_dict)
if step % 100 == 0:
[loss_val, summary_str] = sess.run([total_loss, summary_op], feed_dict=feed_dict)
print "%s Step %d: Training loss %f" % (datetime.now(), step, loss_val)
summary_writer.add_summary(summary_str, global_step=step)
saver.save(sess, FLAGS.logs_dir + "model.ckpt", global_step=step)
def inference(image, keep_prob):
"""
Semantic segmentation network definition
:param image: input image. Should have values in range 0-255
:param keep_prob:
:return:
"""
print("setting up vgg initialized conv layers ...")
model_data = utils.get_model_data(FLAGS.model_dir, MODEL_URL)
mean = model_data['normalization'][0][0][0]
mean_pixel = np.mean(mean, axis=(0, 1))
weights = np.squeeze(model_data['layers'])
#processed_image = utils.process_image(image, mean_pixel)
with tf.variable_scope("inference"):
image_net = vgg_net(weights, image)
conv_final_layer = image_net["conv5_3"]
pool5 = utils.max_pool_2x2(conv_final_layer)
W6 = utils.weight_variable([7, 7, 512, 4096], name="W6")
b6 = utils.bias_variable([4096], name="b6")
conv6 = utils.conv2d_basic(pool5, W6, b6)
relu6 = tf.nn.relu(conv6, name="relu6")
if FLAGS.debug:
utils.add_activation_summary(relu6)
relu_dropout6 = tf.nn.dropout(relu6, keep_prob=keep_prob)
W7 = utils.weight_variable([1, 1, 4096, 4096], name="W7")
b7 = utils.bias_variable([4096], name="b7")
conv7 = utils.conv2d_basic(relu_dropout6, W7, b7)
relu7 = tf.nn.relu(conv7, name="relu7")
if FLAGS.debug:
utils.add_activation_summary(relu7)
relu_dropout7 = tf.nn.dropout(relu7, keep_prob=keep_prob)
W8 = utils.weight_variable([1, 1, 4096, NUM_OF_CLASSESS], name="W8")
b8 = utils.bias_variable([NUM_OF_CLASSESS], name="b8")
conv8 = utils.conv2d_basic(relu_dropout7, W8, b8)
# annotation_pred1 = tf.argmax(conv8, dimension=3, name="prediction1")
# now to upscale to actual image size
deconv_shape1 = image_net["pool4"].get_shape()
W_t1 = utils.weight_variable([4, 4, deconv_shape1[3].value, NUM_OF_CLASSESS], name="W_t1")
b_t1 = utils.bias_variable([deconv_shape1[3].value], name="b_t1")
conv_t1 = utils.conv2d_transpose_strided(conv8, W_t1, b_t1, output_shape=tf.shape(image_net["pool4"]))
fuse_1 = tf.add(conv_t1, image_net["pool4"], name="fuse_1")
deconv_shape2 = image_net["pool3"].get_shape()
W_t2 = utils.weight_variable([4, 4, deconv_shape2[3].value, deconv_shape1[3].value], name="W_t2")
b_t2 = utils.bias_variable([deconv_shape2[3].value], name="b_t2")
conv_t2 = utils.conv2d_transpose_strided(fuse_1, W_t2, b_t2, output_shape=tf.shape(image_net["pool3"]))
fuse_2 = tf.add(conv_t2, image_net["pool3"], name="fuse_2")
shape = tf.shape(image)
deconv_shape3 = tf.stack([shape[0], shape[1], shape[2], NUM_OF_CLASSESS])
W_t3 = utils.weight_variable([16, 16, NUM_OF_CLASSESS, deconv_shape2[3].value], name="W_t3")
b_t3 = utils.bias_variable([NUM_OF_CLASSESS], name="b_t3")
conv_t3 = utils.conv2d_transpose_strided(fuse_2, W_t3, b_t3, output_shape=deconv_shape3, stride=8)
annotation_pred = tf.argmax(conv_t3, dimension=3, name="prediction")
return tf.expand_dims(annotation_pred, dim=3), conv_t3
def mlp(inputs, output_dimension, scope=""):
shape = inputs.get_shape().as_list()
W_fc1 = utils.weight_variable([shape[1], output_dimension])
b_fc1 = utils.bias_variable([output_dimension])
linear = tf.matmul(inputs, W_fc1) + b_fc1
return linear
