【收藏】简单易用 TensorFlow 代码集，GAN通用框架、函数

新智元报道

来源：GitHub 作者：Junho Kim

【新智元导读】今天为大家推荐一个实用的GitHub项目：TensorFlow-Cookbook。这是一个易用的TensorFlow代码集，包含了对GAN有用的一些通用架构和函数。

今天为大家推荐一个实用的GitHub项目：TensorFlow-Cookbook。

这是一个易用的TensorFlow代码集，作者是来自韩国的AI研究科学家Junho Kim，内容涵盖了谱归一化卷积、部分卷积、pixel shuffle、几种归一化函数、 tf-datasetAPI，等等。

作者表示，这个repo包含了对GAN有用的一些通用架构和函数。

项目正在进行中，作者将持续为其他领域添加有用的代码，目前正在添加的是 tf-Eager mode的代码。欢迎提交pull requests和issues。

Github地址 :

https://github.com/taki0112/Tensorflow-Cookbook

如何使用

Import

ops.py
- operations
- from ops import *
utils.py
- image processing
- from utils import *

Network template

def network(x, is_training=True, reuse=False, scope="network"):    with tf.variable_scope(scope, reuse=reuse):
        x = conv(...)        
        ...
        
        return logit

使用DatasetAPI向网络插入数据

Image_Data_Class = ImageData(img_size, img_ch, augment_flag)
trainA = trainA.map(Image_Data_Class.image_processing, num_parallel_calls=16)
trainA = trainA.shuffle(buffer_size=10000).prefetch(buffer_size=batch_size).batch(batch_size).repeat()

trainA_iterator = trainA.make_one_shot_iterator()
data_A = trainA_iterator.get_next()
logit = network(data_A)

了解更多，请阅读： https://github.com/taki0112/Tensorflow-DatasetAPI

Option

padding='SAME'
- pad = ceil[ (kernel - stride) / 2 ]
pad_type
- 'zero' or 'reflect'
sn
- use spectral_normalization or not
Ra
- use relativistic gan or not
loss_func
- gan
- lsgan
- hinge
- wgan
- wgan-gp
- dragan

注意

如果你不想共享变量，请以不同的方式设置所有作用域名称。

权重(Weight)

weight_init = tf.truncated_normal_initializer(mean=0.0, stddev=0.02)
weight_regularizer = tf.contrib.layers.l2_regularizer(0.0001)
weight_regularizer_fully = tf.contrib.layers.l2_regularizer(0.0001)

初始化(Initialization)

Xavier : tf.contrib.layers.xavier_initializer()
He : tf.contrib.layers.variance_scaling_initializer()
Normal : tf.random_normal_initializer(mean=0.0, stddev=0.02)
Truncated_normal : tf.truncated_normal_initializer(mean=0.0, stddev=0.02)
Orthogonal : tf.orthogonal_initializer(1.0) / # if relu = sqrt(2), the others = 1.0

正则化(Regularization)

l2_decay : tf.contrib.layers.l2_regularizer(0.0001)
orthogonal_regularizer : orthogonal_regularizer(0.0001) & orthogonal_regularizer_fully(0.0001)

卷积(Convolution)

basic conv

x = conv(x, channels=64, kernel=3, stride=2, pad=1, pad_type='reflect', use_bias=True, sn=True, scope='conv')

partial conv (NVIDIA Partial Convolution)

x = partial_conv(x, channels=64, kernel=3, stride=2, use_bias=True, padding='SAME', sn=True, scope='partial_conv')

dilated conv

x = dilate_conv(x, channels=64, kernel=3, rate=2, use_bias=True, padding='SAME', sn=True, scope='dilate_conv')

Deconvolution

basic deconv

x = deconv(x, channels=64, kernel=3, stride=2, padding='SAME', use_bias=True, sn=True, scope='deconv')

Fully-connected

x = fully_conneted(x, units=64, use_bias=True, sn=True, scope='fully_connected')

Pixel shuffle

x = conv_pixel_shuffle_down(x, scale_factor=2, use_bias=True, sn=True, scope='pixel_shuffle_down')
x = conv_pixel_shuffle_up(x, scale_factor=2, use_bias=True, sn=True, scope='pixel_shuffle_up')

down ===> [height, width] -> [height // scale_factor, width // scale_factor]
up ===> [height, width] -> [height * scale_factor, width * scale_factor]

Block

residual block

x = resblock(x, channels=64, is_training=is_training, use_bias=True, sn=True, scope='residual_block')
x = resblock_down(x, channels=64, is_training=is_training, use_bias=True, sn=True, scope='residual_block_down')
x = resblock_up(x, channels=64, is_training=is_training, use_bias=True, sn=True, scope='residual_block_up')

down ===> [height, width] -> [height // 2, width // 2]
up ===> [height, width] -> [height * 2, width * 2]

attention block

x = self_attention(x, channels=64, use_bias=True, sn=True, scope='self_attention')
x = self_attention_with_pooling(x, channels=64, use_bias=True, sn=True, scope='self_attention_version_2')

x = squeeze_excitation(x, channels=64, ratio=16, use_bias=True, sn=True, scope='squeeze_excitation')

x = convolution_block_attention(x, channels=64, ratio=16, use_bias=True, sn=True, scope='convolution_block_attention')

Normalization

Normalization

x = batch_norm(x, is_training=is_training, scope='batch_norm')
x = instance_norm(x, scope='instance_norm')
x = layer_norm(x, scope='layer_norm')
x = group_norm(x, groups=32, scope='group_norm')

x = pixel_norm(x)

x = batch_instance_norm(x, scope='batch_instance_norm')

x = condition_batch_norm(x, z, is_training=is_training, scope='condition_batch_norm'):

x = adaptive_instance_norm(x, gamma, beta):

如何使用 condition_batch_norm，请参考：

https://github.com/taki0112/BigGAN-Tensorflow

如何使用 adaptive_instance_norm，请参考：

https://github.com/taki0112/MUNIT-Tensorflow

Activation

x = relu(x)
x = lrelu(x, alpha=0.2)
x = tanh(x)
x = sigmoid(x)
x = swish(x)

Pooling & Resize

x = up_sample(x, scale_factor=2)

x = max_pooling(x, pool_size=2)
x = avg_pooling(x, pool_size=2)

x = global_max_pooling(x)
x = global_avg_pooling(x)

x = flatten(x)
x = hw_flatten(x)

Loss

classification loss

loss, accuracy = classification_loss(logit, label)

pixel loss

loss = L1_loss(x, y)
loss = L2_loss(x, y)
loss = huber_loss(x, y)
loss = histogram_loss(x, y)

histogram_loss 表示图像像素值在颜色分布上的差异。

gan loss

d_loss = discriminator_loss(Ra=True, loss_func='wgan-gp', real=real_logit, fake=fake_logit)
g_loss = generator_loss(Ra=True, loss_func='wgan_gp', real=real_logit, fake=fake_logit)

如何使用 gradient_penalty，请参考：

https://github.com/taki0112/BigGAN-Tensorflow/blob/master/BigGAN_512.py#L180

kl-divergence (z ~ N(0, 1))

loss = kl_loss(mean, logvar)

Author

Junho Kim

Github地址 :