some doc change

CHANGES.md

@@ -8,6 +8,8 @@ so you won't need to look at here very often.
 Here are a list of things that were changed, starting from an early version.
 TensorFlow itself also changes API and those are not listed here.
 
++ [2017/04/09](https://github.com/ppwwyyxx/tensorpack/commit/5beab907895aec36bdcaed62e25b976aad7979b8).
+	`ParamRestore` was renamed to `DictRestore`.
 + [2017/03/16](https://github.com/ppwwyyxx/tensorpack/commit/ccae46f4a3ca89dc3df901a338eef8447d19a730).
 	`session_config` option in `TrainConfig` and `PredictConfig` is deprecated.
 	Use `session_creator` to define how to create session instead.

docs/tutorial/extend/augmentor.md

+
+### Write an image augmentor
+
+First thing to note: an augmentor is a part of the DataFlow, so you can always
+[write a DataFlow](http://tensorpack.readthedocs.io/en/latest/tutorial/extend/dataflow.html)
+to do whatever operations to your data, rather than writing an augmentor.
+Augmentors just sometimes make things easier.
+
+An augmentor maps images to images.
+If you have such a mapping function `f` already, you can simply use `imgaug.MapImage(f)` as the
+augmentor, or use `MapDataComponent(df, f, index)` as the DataFlow.
+In other words, for simple mapping you don't need to write an augmentor.
+
+An augmentor does something more than applying the mapping. The interface you'll need to implement
+is:
+
+```python
+class MyAug(imgaug.ImageAugmentor):
+	def _get_augment_params(self, img):
+		# generated random params with self.rng
+	  return params
+
+  def _augment(self, img, params):
+	  return augmented_img
+```
+
+It does the following extra things for you:
+
+1. `self.rng` is a `np.random.RandomState` object,
+	guranteed to have different seeds when you use multiprocess prefetch.
+	In multiprocess settings, you'll always need it to generate random numbers.
+
+2. Random parameters and the actual augmentation is separated. This allows you to apply the
+	same random transformation to several images (with `AugmentImageComponents`),
+	which is important to tasks such as segmentation.

docs/tutorial/faq.md

@@ -3,14 +3,14 @@
 
 ## Does it support data format X / augmentation Y / layer Z?
 
-The library tries to support everything, but it couldn't really include everything.
+The library tries to __support__ everything, but it couldn't really __include__ everything.
 
-For the XYZ you need, you can either implement them, or use any existing code and wrap it
+For your XYZ, you can either implement them, or use any existing python code and wrap it
 with tensorpack interface. See [Extend Tensorpack](http://tensorpack.readthedocs.io/en/latest/tutorial/index.html#extend-tensorpack)
 for more details.
 
-It you think:
-1. The framework has limitation so your XYZ cannot be supported, OR
+If you think:
+1. The framework has limitation in its interface so your XYZ cannot be supported, OR
 2. Your XYZ is very common, or very well-defined, so it would be nice to include it.
 
 Then it's a good time to open an issue.
@@ -21,26 +21,27 @@ When you enable `ModelSaver` as a callback,
 trained models will be stored in TensorFlow checkpoint format, which typically includes a
 `.data-xxxxx` file and a `.index` file. Both are necessary.
 
-To inspect a checkpoint, the easiest way is `tf.train.NewCheckpointReader`. Please note that it
-expects a path without the extension.
+To inspect a checkpoint, the easiest tool is `tf.train.NewCheckpointReader`. Please note that it
+expects a model path without the extension.
 
-You can dump a cleaner version of the model (with only model/trainable variables), with
+You can dump a cleaner version of the model (without unnecessary variables), using
 `scripts/dump-model-params.py`, as a simple `var-name: value` dict saved in npy format.
 The script expects a metagraph file which is also saved by `ModelSaver`.
 
 
 ## How to load a model / do transfer learning
 
-All model loading (in either training or testing) is through the `session_init` interface
+All model loading (in either training or testing) is through the `session_init` option
 in `TrainConfig` or `PredictConfig`.
 It accepts a `SessionInit` instance, where the common options are `SaverRestore` which restores
 TF checkpoint, or `DictRestore` which restores a dict. `get_model_loader` is a small helper to
 decide which one to use from file name.
 
-Doing transfer learning is painless. Variable restoring is completely based on name match between
+Doing transfer learning is straightforward. Variable restoring is completely based on name match between
 the current graph and the `SessionInit` initializer.
-Therefore, if you want to re-train some layer, just rename it.
-And unmatched variables on both side will be printed as warning.
+Therefore, if you want to load some model, just use the same name.
+If you want to re-train some layer, just rename it.
+Unmatched variables on both side will be printed as warning.
 
 To freeze some variables, there are [different ways](https://github.com/ppwwyyxx/tensorpack/issues/87#issuecomment-270545291)
 with pros and cons.

docs/tutorial/index.rst

@@ -48,6 +48,7 @@ Extend Tensorpack
   :maxdepth: 1
 
   extend/dataflow
+  extend/augmentor
   extend/model
   extend/trainer
   extend/callback

examples/GAN/WGAN.py

@@ -18,13 +18,8 @@ See the docstring in DCGAN.py for usage.
 """
 
 # Don't want to mix two examples together, but want to reuse the code.
-# So here just import stuff from DCGAN-CelebA, and change the batch size & model
-import imp
-DCGAN = imp.load_source(
-    'DCGAN',
-    os.path.join(os.path.dirname(__file__), 'DCGAN.py'))
-
-
+# So here just import stuff from DCGAN, and change the batch size & model
+import DCGAN
 G.BATCH = 64
 
 

tensorpack/dataflow/imgaug/base.py

@@ -62,20 +62,6 @@ class Augmentor(object):
 
 
 class ImageAugmentor(Augmentor):
-
-    def augment(self, img):
-        """
-        Perform augmentation on the image (possibly) in-place.
-
-        Args:
-            img (np.ndarray): an [h,w] or [h,w,c] image.
-
-        Returns:
-            np.ndarray: the augmented image, always of type float32.
-        """
-        img, params = self._augment_return_params(img)
-        return img
-
     def _fprop_coord(self, coord, param):
         return coord