add a simple conditional gan

CHANGES.md

@@ -19,6 +19,7 @@ TensorFlow itself also changes API and those are not listed here.
 	See [commit](https://github.com/ppwwyyxx/tensorpack/commit/651a5aea8f9aacad7147542021dcf106fc824bc2) to change your code.
 * 2016/12/15. The `predict_tower` option is in `TrainConfig` now instead of `Trainer`. See
 	[commit](https://github.com/ppwwyyxx/tensorpack/commit/99c70935a7f72050f45891fbbcc49c4ce43aedce).
+* 2016/11/10. The `{input,output}_var_names` argument in `PredictConfig` is renamed to `{input,output}_names`. See [commit](https://github.com/ppwwyyxx/tensorpack/commit/77bcc8b1afc984a569f6ec3eda0a3c47b4e2923a).
 * 2016/11/06. The inferencer `ClassificationError` now expects the vector tensor returned by
 	`prediction_incorrect` instead of the "wrong" tensor. See [commit](https://github.com/ppwwyyxx/tensorpack/commit/740e9d8ca146af5a911f68a369dd7348243a2253)
 	to make changes.

examples/GAN/ConditionalGAN-mnist.py

+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# File: ConditionalGAN-mnist.py
+# Author: Yuxin Wu <ppwwyyxxc@gmail.com>
+
+import numpy as np
+import tensorflow as tf
+import os
+import sys
+import cv2
+import argparse
+
+from tensorpack import *
+from tensorpack.utils.viz import *
+import tensorpack.tfutils.symbolic_functions as symbf
+from GAN import GANTrainer, RandomZData, GANModelDesc
+
+"""
+To train:
+    ./ConditionalGAN-mnist.py
+
+To visualize:
+    ./ConditionalGAN-mnist.py --sample --load path/to/model
+"""
+
+BATCH = 128
+
+
+class Model(GANModelDesc):
+    def _get_input_vars(self):
+        return [InputVar(tf.float32, (None, 28, 28), 'input'),
+                InputVar(tf.int32, (None,), 'label')]
+
+    def generator(self, z, y):
+        l = FullyConnected('fc0', tf.concat([z, y], 1), 1024, nl=BNReLU)
+        l = FullyConnected('fc1', tf.concat([l, y], 1), 64 * 2 * 7 * 7, nl=BNReLU)
+        l = tf.reshape(l, [-1, 7, 7, 64 * 2])
+
+        y = tf.reshape(y, [-1, 1, 1, 10])
+        l = tf.concat([l, tf.tile(y, [1, 7, 7, 1])], 3)
+        l = Deconv2D('deconv1', l, [14, 14, 64 * 2], 5, 2, nl=BNReLU)
+
+        l = tf.concat([l, tf.tile(y, [1, 14, 14, 1])], 3)
+        l = Deconv2D('deconv2', l, [28, 28, 1], 5, 2, nl=tf.identity)
+        l = tf.nn.tanh(l, name='gen')
+        return l
+
+    def discriminator(self, imgs, y):
+        """ return a (b, 1) logits"""
+        yv = y
+        y = tf.reshape(y, [-1, 1, 1, 10])
+        with argscope(Conv2D, nl=tf.identity, kernel_shape=5, stride=2), \
+                argscope(LeakyReLU, alpha=0.2):
+            l = (LinearWrap(imgs)
+                 .ConcatWith(tf.tile(y, [1, 28, 28, 1]), 3)
+                 .Conv2D('conv0', 11)
+                 .LeakyReLU()
+
+                 .ConcatWith(tf.tile(y, [1, 14, 14, 1]), 3)
+                 .Conv2D('conv1', 74)
+                 .BatchNorm('bn1').LeakyReLU()
+
+                 .apply(symbf.batch_flatten)
+                 .ConcatWith(yv, 1)
+                 .FullyConnected('fc1', 1024, nl=tf.identity)
+                 .BatchNorm('bn2').LeakyReLU()
+
+                 .ConcatWith(yv, 1)
+                 .FullyConnected('fct', 1, nl=tf.identity)())
+        return l
+
+    def _build_graph(self, input_vars):
+        image_pos, y = input_vars
+        image_pos = tf.expand_dims(image_pos * 2.0 - 1, -1)
+        y = tf.one_hot(y, 10, name='label_onehot')
+
+        z = tf.random_uniform([BATCH, 100], -1, 1, name='z_train')
+        z = symbf.shapeless_placeholder(z, [0], name='z')
+
+        with argscope([Conv2D, Deconv2D, FullyConnected],
+                      W_init=tf.truncated_normal_initializer(stddev=0.02)):
+            with tf.variable_scope('gen'):
+                image_gen = self.generator(z, y)
+                tf.summary.image('gen', image_gen, 30)
+            with tf.variable_scope('discrim'):
+                vecpos = self.discriminator(image_pos, y)
+            with tf.variable_scope('discrim', reuse=True):
+                vecneg = self.discriminator(image_gen, y)
+
+        self.build_losses(vecpos, vecneg)
+        self.collect_variables()
+
+
+def get_data():
+    ds = ConcatData([dataset.Mnist('train'), dataset.Mnist('test')])
+    ds = BatchData(ds, BATCH)
+    return ds
+
+
+def get_config():
+    logger.auto_set_dir()
+    dataset = get_data()
+    lr = symbf.get_scalar_var('learning_rate', 2e-4, summary=True)
+    return TrainConfig(
+        dataflow=dataset,
+        optimizer=tf.train.AdamOptimizer(lr, beta1=0.5, epsilon=1e-3),
+        callbacks=[ModelSaver()],
+        session_config=get_default_sess_config(0.5),
+        model=Model(),
+        steps_per_epoch=500,
+        max_epoch=100,
+    )
+
+
+def sample(model_path):
+    pred = PredictConfig(
+        session_init=get_model_loader(model_path),
+        model=Model(),
+        input_names=['label', 'z'],
+        output_names=['gen/gen'])
+
+    ds = MapData(RandomZData((100, 100)),
+                 lambda dp: [np.arange(100) % 10, dp[0]])
+    pred = SimpleDatasetPredictor(pred, ds)
+    for o in pred.get_result():
+        o = o[0] * 255.0
+        viz = next(build_patch_list(o, nr_row=10, nr_col=10))
+        viz = cv2.resize(viz, (800, 800))
+        interactive_imshow(viz)
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--gpu', help='comma separated list of GPU(s) to use.')
+    parser.add_argument('--load', help='load model')
+    parser.add_argument('--sample', action='store_true')
+    args = parser.parse_args()
+    if args.gpu:
+        os.environ['CUDA_VISIBLE_DEVICES'] = args.gpu
+    if args.sample:
+        sample(args.load)
+    else:
+        config = get_config()
+        if args.load:
+            config.session_init = SaverRestore(args.load)
+        GANTrainer(config).train()

examples/GAN/DCGAN-CelebA.py

@@ -27,7 +27,7 @@ The original code (dcgan.torch) uses kernel_shape=4, but I found the difference
 2. Start training:
     ./DCGAN-CelebA.py --data /path/to/image_align_celeba/
 3. Visualize samples of a trained model:
-    ./DCGAN-CelebA.py --load model.tfmodel --sample
+    ./DCGAN-CelebA.py --load path/to/model --sample
 """
 
 CFG.SHAPE = 64

examples/GAN/InfoGAN-mnist.py

@@ -17,6 +17,14 @@ import tensorpack.tfutils.symbolic_functions as symbf
 from tensorpack.tfutils.gradproc import ScaleGradient, CheckGradient
 from GAN import GANTrainer, GANModelDesc
 
+"""
+To train:
+    ./InfoGAN-mnist.py
+
+To visualize:
+    ./InfoGAN-mnist.py --sample --load path/to/model
+"""
+
 BATCH = 128
 NOISE_DIM = 62
 

examples/GAN/README.md

 # Generative Adversarial Networks
 
-Reproduce the following GAN-related papers:
+Reproduce the following GAN-related methods:
 
-+ Unsupervised Representation Learning with DCGAN. [paper](https://arxiv.org/abs/1511.06434)
++ DCGAN ([Unsupervised Representation Learning with DCGAN](https://arxiv.org/abs/1511.06434))
 
-+ Image-to-image Translation with Conditional Adversarial Networks. [paper](https://arxiv.org/pdf/1611.07004v1.pdf)
++ pix2pix ([Image-to-image Translation with Conditional Adversarial Networks](https://arxiv.org/pdf/1611.07004v1.pdf))
 
-+ InfoGAN: Interpretable Representation Learning by Information Maximizing GAN. [paper](https://arxiv.org/abs/1606.03657)
++ InfoGAN ([InfoGAN: Interpretable Representation Learning by Information Maximizing GAN](https://arxiv.org/abs/1606.03657))
+
++ Conditional GAN
 
 Please see the __docstring__ in each script for detailed usage.
 
@@ -46,3 +48,6 @@ It then maximizes mutual information between these latent variables and the imag
 * Middle: 1 continuous latent variable controlled the rotation.
 * Right: another continuous latent variable controlled the thickness.
 
+## ConditionalGAN-mnist.py
+
+Train a simple GAN on mnist, conditioned on the class labels.