Merge pull request #19 from tensorlayer/tl2

Tl2

Author: zsdonghao

README.md

@@ -1,31 +1,27 @@
-# DCGAN in TensorFlow
+# DCGAN in TensorLayer
 
-TensorFlow / TensorLayer implementation of [Deep Convolutional Generative Adversarial Networks](http://arxiv.org/abs/1511.06434) which is a stabilize Generative Adversarial Networks.
+TensorLayer implementation of [Deep Convolutional Generative Adversarial Networks](http://arxiv.org/abs/1511.06434).
 
 Looking for Text to Image Synthesis ? [click here](https://github.com/zsdonghao/text-to-image)
 
 ![alt tag](img/DCGAN.png)
 
-* [Brandon Amos](http://bamos.github.io/) wrote an excellent [blog post](http://bamos.github.io/2016/08/09/deep-completion/) and [image completion code](https://github.com/bamos/dcgan-completion.tensorflow) based on this repo.
-* *To avoid the fast convergence of D (discriminator) network, G (generator) network is updated twice for each D network update, which differs from original paper.*
-
-
 ## Prerequisites
 
-- Python 2.7 or Python 3.3+
-- [TensorFlow==1.10.0+](https://www.tensorflow.org/)
-- [TensorLayer==1.10.1+](https://github.com/tensorlayer/tensorlayer)
-
+- Python3
+- TensorFlow==1.13
+- TensorLayer (self-contained)
 
 ## Usage
 
-First, download images to `data/celebA`:
-
-    $ python download.py celebA		[202599 face images]
+First, download the aligned face images from [google](https://drive.google.com/open?id=0B7EVK8r0v71pWEZsZE9oNnFzTm8) or [baidu](https://pan.baidu.com/s/1eSNpdRG#list/path=%2F) to a `data` folder.
 
 Second, train the GAN:
 
-    $ python main.py
+    $ python main_eager_mode.py   # single GPU for TF>=1.13
+    $ python main_graph_mode.py   # single GPU for TF<=1.13
+    $ python main_eager_mode_horovod.py   # multiple GPU (TODO)
+    $ python main_eager_mode_tlmagic.py   # multiple GPU (TODO)    
 
 ## Result on celebA
 

download.py renamed to _backup/download.py

@@ -0,0 +1,31 @@
+"""
+Deep learning and Reinforcement learning library for Researchers and Engineers
+"""
+# from __future__ import absolute_import
+
+
+try:
+    install_instr = "Please make sure you install a recent enough version of TensorFlow."
+    import tensorflow
+except ImportError:
+    raise ImportError("__init__.py : Could not import TensorFlow." + install_instr)
+
+from . import activation
+act = activation
+from . import cost
+from . import files
+# from . import init
+from . import iterate
+from . import layers
+from . import ops
+from . import utils
+from . import visualize
+from . import prepro        # was preprocesse
+from . import nlp
+from . import rein
+
+
+__version__ = "1.4.2"
+
+global_flag = {}
+global_dict = {}

_tensorlayer/__init__.py

@@ -0,0 +1,109 @@
+#! /usr/bin/python
+# -*- coding: utf8 -*-
+
+
+
+import tensorflow as tf
+
+def identity(x, name=None):
+    """The identity activation function, Shortcut is ``linear``.
+
+    Parameters
+    ----------
+    x : a tensor input
+        input(s)
+
+
+    Returns
+    --------
+    A `Tensor` with the same type as `x`.
+    """
+    return x
+
+# Shortcut
+linear = identity
+
+def ramp(x=None, v_min=0, v_max=1, name=None):
+    """The ramp activation function.
+
+    Parameters
+    ----------
+    x : a tensor input
+        input(s)
+    v_min : float
+        if input(s) smaller than v_min, change inputs to v_min
+    v_max : float
+        if input(s) greater than v_max, change inputs to v_max
+    name : a string or None
+        An optional name to attach to this activation function.
+
+
+    Returns
+    --------
+    A `Tensor` with the same type as `x`.
+    """
+    return tf.clip_by_value(x, clip_value_min=v_min, clip_value_max=v_max, name=name)
+
+def leaky_relu(x=None, alpha=0.1, name="LeakyReLU"):
+    """The LeakyReLU, Shortcut is ``lrelu``.
+
+    Modified version of ReLU, introducing a nonzero gradient for negative
+    input.
+
+    Parameters
+    ----------
+    x : A `Tensor` with type `float`, `double`, `int32`, `int64`, `uint8`,
+        `int16`, or `int8`.
+    alpha : `float`. slope.
+    name : a string or None
+        An optional name to attach to this activation function.
+
+    Examples
+    ---------
+    >>> network = tl.layers.DenseLayer(network, n_units=100, name = 'dense_lrelu',
+    ...                 act= lambda x : tl.act.lrelu(x, 0.2))
+
+    References
+    ------------
+    - `Rectifier Nonlinearities Improve Neural Network Acoustic Models, Maas et al. (2013) <http://web.stanford.edu/~awni/papers/relu_hybrid_icml2013_final.pdf>`_
+    """
+    with tf.name_scope(name) as scope:
+        # x = tf.nn.relu(x)
+        # m_x = tf.nn.relu(-x)
+        # x -= alpha * m_x
+        x = tf.maximum(x, alpha * x)
+    return x
+
+#Shortcut
+lrelu = leaky_relu
+
+def pixel_wise_softmax(output, name='pixel_wise_softmax'):
+    """Return the softmax outputs of images, every pixels have multiple label, the sum of a pixel is 1.
+    Usually be used for image segmentation.
+
+    Parameters
+    ------------
+    output : tensor
+        - For 2d image, 4D tensor [batch_size, height, weight, channel], channel >= 2.
+        - For 3d image, 5D tensor [batch_size, depth, height, weight, channel], channel >= 2.
+
+    Examples
+    ---------
+    >>> outputs = pixel_wise_softmax(network.outputs)
+    >>> dice_loss = 1 - dice_coe(outputs, y_, epsilon=1e-5)
+
+    References
+    -----------
+    - `tf.reverse <https://www.tensorflow.org/versions/master/api_docs/python/array_ops.html#reverse>`_
+    """
+    with tf.name_scope(name) as scope:
+        return tf.nn.softmax(output)
+        ## old implementation
+        # exp_map = tf.exp(output)
+        # if output.get_shape().ndims == 4:   # 2d image
+        #     evidence = tf.add(exp_map, tf.reverse(exp_map, [False, False, False, True]))
+        # elif output.get_shape().ndims == 5: # 3d image
+        #     evidence = tf.add(exp_map, tf.reverse(exp_map, [False, False, False, False, True]))
+        # else:
+        #     raise Exception("output parameters should be 2d or 3d image, not %s" % str(output._shape))
+        # return tf.div(exp_map, evidence)