v1.0.1

.gitignore

@@ -3,3 +3,4 @@
 *.pyc
 build/
 data/
+docs/html/

README.md

@@ -13,6 +13,7 @@ year = {2017}
 ```
 
 ## [Documentation](http://hangzh.com/PyTorch-Encoding/)
-Please visit the [**Docs**](http://hangzh.com/PyTorch-Encoding/) for detail instructions of installation and usage.
 
-(If you would like to reproduce the texture recognition benchmark in the paper, please visit our original [Torch implementation](https://github.com/zhanghang1989/Deep-Encoding).)
+- Please visit the [**Docs**](http://hangzh.com/PyTorch-Encoding/) for detail instructions of installation and usage. 
+
+- [**Link**](http://hangzh.com/PyTorch-Encoding/experiments/texture.html) to the experiments and pre-trained models.

build.py

@@ -40,9 +40,9 @@ defines = [('WITH_CUDA', None)]
 with_cuda = True 
 
 include_path = [os.path.join(lib_path, 'include'),
-                                os.path.join(os.environ['HOME'],'pytorch/torch/lib/THC'), 
-                                os.path.join(lib_path,'include/ENCODING'), 
-                                os.path.join(this_file,'encoding/src/')]
+                os.path.join(os.environ['HOME'],'pytorch/torch/lib/THC'), 
+                os.path.join(lib_path,'include/ENCODING'), 
+                os.path.join(this_file,'encoding/src/')]
 
 def make_relative_rpath(path):
     if platform.system() == 'Darwin':

docs/source/_static/css/encoding.css

+body {
+    font-family: "Lato","proxima-nova","Helvetica Neue",Arial,sans-serif;
+}
+
+/* Default header fonts are ugly */
+h1, h2, .rst-content .toctree-wrapper p.caption, h3, h4, h5, h6, legend, p.caption {
+    font-family: "Lato","proxima-nova","Helvetica Neue",Arial,sans-serif;
+}
+
+/* Use white for docs background */
+.wy-side-nav-search {
+    background-color: #a0e2ff;
+}
+
+.wy-nav-content-wrap, .wy-menu li.current > a  {
+    background-color: #fff;
+}
+
+@media screen and (min-width: 1400px) {
+    .wy-nav-content-wrap {
+        background-color: rgba(0, 0, 0, 0.0470588);
+    }
+
+    .wy-nav-content {
+        background-color: #fff;
+    }
+}
+
+/* Fixes for mobile */
+.wy-nav-top {
+    background-color: #fff;
+    background-repeat: no-repeat;
+    background-position: center;
+    padding: 0;
+    margin: 0.4045em 0.809em;
+    color: #333;
+}
+
+.wy-nav-top > a {
+    display: none;
+}
+
+@media screen and (max-width: 768px) {
+    .wy-side-nav-search>a img.logo {
+        height: 60px;
+    }
+}
+
+/* This is needed to ensure that logo above search scales properly */
+.wy-side-nav-search a {
+    display: block;
+}
+
+/* This ensures that multiple constructors will remain in separate lines. */
+.rst-content dl:not(.docutils) dt {
+    display: table;
+}
+
+/* Use our blue for literals */
+.rst-content tt.literal, .rst-content tt.literal, .rst-content code.literal {
+    color: #4080bf;
+}
+
+.rst-content tt.xref, a .rst-content tt, .rst-content tt.xref,
+.rst-content code.xref, a .rst-content tt, a .rst-content code {
+    color: #404040;
+}
+
+/* Change link colors (except for the menu) */
+
+a {
+    color: #4080bf;
+}
+
+a:hover {
+    color: #4080bf;
+}
+
+
+a:visited {
+    color: #306293;
+}
+
+.wy-menu a {
+    color: #b3b3b3;
+}
+
+.wy-menu a:hover {
+    color: #b3b3b3;
+}
+
+/* Default footer text is quite big */
+footer {
+    font-size: 80%;
+}
+
+footer .rst-footer-buttons {
+    font-size: 125%; /* revert footer settings - 1/80% = 125% */
+}
+
+footer p {
+    font-size: 100%;
+}
+
+/* For hidden headers that appear in TOC tree */
+/* see http://stackoverflow.com/a/32363545/3343043 
+*/
+.rst-content .hidden-section {
+    display: none;
+}
+
+nav .hidden-section {
+    display: inherit;
+}
+
+.wy-side-nav-search>div.version {
+    color: #000;
+}

docs/source/conf.py

@@ -20,6 +20,7 @@
 # import os
 # import sys
 # sys.path.insert(0, os.path.abspath('.'))
+import encoding
 import sphinx_rtd_theme
 
 
@@ -47,7 +48,7 @@ extensions = [
 
 napoleon_use_ivar = True
 
-googleanalytics_id = 'UA-90545585-1'
+googleanalytics_id = 'UA-54746507-1'
 googleanalytics_enabled = True
 
 # Add any paths that contain templates here, relative to this directory.
@@ -56,8 +57,8 @@ templates_path = ['_templates']
 # The suffix(es) of source filenames.
 # You can specify multiple suffix as a list of string:
 #
-# source_suffix = ['.rst', '.md']
-source_suffix = '.rst'
+source_suffix = ['.rst', '.md']
+#source_suffix = '.rst'
 
 # The master toctree document.
 master_doc = 'index'
@@ -72,7 +73,7 @@ author = 'Hang Zhang'
 # built documents.
 #
 # The short X.Y version.
-version = 'master (0.0.1)'
+version = 'master (' + encoding.__version__ + ')'
 # The full version, including alpha/beta/rc tags.
 # TODO: verify this works as expected
 release = 'master'
@@ -124,6 +125,7 @@ html_static_path = ['_static']
 html_context = {
     'css_files': [
         'https://fonts.googleapis.com/css?family=Lato',
+        '_static/css/encoding.css'
     ],
 }
 #'_static/css/hangzh.css'

docs/source/dilated.rst

+.. role:: hidden
+    :class: hidden-section
+
+Dilated Networks
+================
+
+We provide correct dilated pre-trained ResNet and DenseNet for semantic segmentation. 
+For dilation of ResNet, we replace the stride of 2 Conv3x3 at begining of certain stage and update the dilation of the conv layers afterwards. 
+For dilation of DenseNet, we provide DilatedAvgPool2d that handles the dilation of the transition layers, then update the dilation of the conv layers afterwards. 
+All provided models have been verified. 
+
+
+.. automodule:: encoding.dilated
+.. currentmodule:: encoding.dilated
+
+ResNet
+------
+
+:hidden:`ResNet`
+~~~~~~~~~~~~~~~~
+
+.. autoclass:: ResNet
+    :members:
+
+:hidden:`resnet18`
+~~~~~~~~~~~~~~~~~~
+
+.. autofunction:: resnet18
+
+:hidden:`resnet34`
+~~~~~~~~~~~~~~~~~~
+
+.. autofunction:: resnet34
+
+:hidden:`resnet50`
+~~~~~~~~~~~~~~~~~~
+
+.. autofunction:: resnet50
+
+:hidden:`resnet101`
+~~~~~~~~~~~~~~~~~~~
+
+.. autofunction:: resnet101
+
+:hidden:`resnet152`
+~~~~~~~~~~~~~~~~~~~
+
+.. autofunction:: resnet152
+
+
+DenseNet
+--------
+
+:hidden:`DenseNet`
+~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: DenseNet
+    :members:
+
+
+:hidden:`densenet161`
+~~~~~~~~~~~~~~~~~~~~~
+
+.. autofunction:: densenet161
+
+
+:hidden:`densenet121`
+~~~~~~~~~~~~~~~~~~~~~
+
+.. autofunction:: densenet121
+
+
+:hidden:`densenet169`
+~~~~~~~~~~~~~~~~~~~~~
+
+.. autofunction:: densenet169
+
+
+:hidden:`densenet201`
+~~~~~~~~~~~~~~~~~~~~~
+
+.. autofunction:: densenet201
+
+

docs/source/encoding.rst

 .. role:: hidden
     :class: hidden-section
 
-Encoding Layer
-==============
-
-.. automodule:: encoding
-
+My NN Layers
+============
 
 Modules
 -------
 
-.. currentmodule:: encoding
+.. currentmodule:: encoding.nn
 
 :hidden:`Encoding`
 ~~~~~~~~~~~~~~~~~~
@@ -18,6 +15,24 @@ Modules
 .. autoclass:: Encoding
     :members:
 
+:hidden:`Inspiration`
+~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: Inspiration
+    :members:
+
+:hidden:`DilatedAvgPool2d`
+~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: DilatedAvgPool2d
+    :members:
+
+:hidden:`GramMatrix`
+~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: GramMatrix
+    :members:
+
 :hidden:`Aggregate`
 ~~~~~~~~~~~~~~~~~~~
 
@@ -27,23 +42,22 @@ Modules
 Functions
 ---------
 
+.. currentmodule:: encoding.functions
+
 :hidden:`aggregate`
 ~~~~~~~~~~~~~~~~~~~
 
-.. autoclass:: aggregate
-    :members:
+.. autofunction:: aggregate
 
 :hidden:`scaledL2`
 ~~~~~~~~~~~~~~~~~~~
 
-.. autoclass:: scaledL2
-    :members:
+.. autofunction:: scaledL2
 
 :hidden:`residual`
 ~~~~~~~~~~~~~~~~~~~
 
-.. autoclass:: residual
-    :members:
+.. autofunction:: residual
 
 
 :hidden:`assign`

docs/source/experiments/style.rst

+MSG-Net Style Transfer Example
+==============================
+
+.. image:: https://raw.githubusercontent.com/zhanghang1989/MSG-Net/master/images/figure1.jpg
+    :width: 55%
+    :align: left
+
+
+We provide PyTorh Implementation of `MSG-Net`_ and `Neural Style`_ in the `GitHub repo <https://github.com/zhanghang1989/PyTorch-Style-Transfer>`_. 
+We also provide `Torch <https://github.com/zhanghang1989/MSG-Net/>`_ and 
+`MXNet <https://github.com/zhanghang1989/MXNet-Gluon-Style-Transfer>`_ implementations.
+
+
+Tabe of content
+---------------
+
+- Real-time Style Transfer using `MSG-Net`_
+    * `Stylize Images using Pre-trained Model`_
+    * `Train Your Own MSG-Net Model`_
+
+- `Neural Style`_
+
+
+MSG-Net
+-------
+
+.. note::
+    Hang Zhang, and Kristin Dana. "Multi-style Generative Network for Real-time Transfer."::
+
+        @article{zhang2017multistyle,
+            title={Multi-style Generative Network for Real-time Transfer},
+            author={Zhang, Hang and Dana, Kristin},
+            journal={arXiv preprint arXiv:1703.06953},
+            year={2017}
+        }
+
+Stylize Images Using Pre-trained Model
+--------------------------------------
+
+- Clone the repo and download the pre-trained model::
+
+    git clone git@github.com:zhanghang1989/PyTorch-Style-Transfer.git
+    cd PyTorch-Style-Transfer/experiments
+    bash models/download_model.sh
+
+- Camera Demo::
+
+    python camera_demo.py demo --model models/9styles.model
+
+.. image:: https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/myimage.gif
+
+- Test the model::
+
+    python main.py eval --content-image images/content/venice-boat.jpg --style-image images/9styles/candy.jpg --model models/9styles.model --content-size 1024
+
+  If you don't have a GPU, simply set ``--cuda=0``. For a different style, set ``--style-image path/to/style``.
+
+  If you would to stylize your own photo, change the ``--content-image path/to/your/photo``. More options:
+
+  * ``--content-image``: path to content image you want to stylize.
+  * ``--style-image``: path to style image (typically covered during the training).
+  * ``--model``: path to the pre-trained model to be used for stylizing the image.
+  * ``--output-image``: path for saving the output image.
+  * ``--content-size``: the content image size to test on.
+  * ``--cuda``: set it to 1 for running on GPU, 0 for CPU.
+
+.. raw:: html
+
+    <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/1.jpg" width="260px" /> <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/2.jpg" width="260px" />
+    <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/3.jpg" width="260px" />
+    <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/4.jpg" width="260px" />
+    <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/5.jpg" width="260px" />
+    <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/6.jpg" width="260px" />
+    <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/7.jpg" width="260px" />
+    <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/8.jpg" width="260px" />
+    <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/9.jpg" width="260px" />
+
+Train Your Own MSG-Net Model
+----------------------------
+
+- Download the dataset::
+  
+    bash dataset/download_dataset.sh
+  
+- Train the model::
+
+    python main.py train --epochs 4
+
+  If you would like to customize styles, set ``--style-folder path/to/your/styles``. More options: 
+
+  * ``--style-folder``: path to the folder style images.
+  * ``--vgg-model-dir``: path to folder where the vgg model will be downloaded. 
+
+  * ``--save-model-dir``: path to folder where trained model will be saved.
+  * ``--cuda``: set it to 1 for running on GPU, 0 for CPU.
+
+
+Neural Style
+-------------
+`Image Style Transfer Using Convolutional Neural Networks <http://www.cv-foundation.org/openaccess/content_cvpr_2016/papers/Gatys_Image_Style_Transfer_CVPR_2016_paper.pdf>`_ by Leon A. Gatys, Alexander S. Ecker, and Matthias Bethge::
+
+    python main.py optim --content-image images/content/venice-boat.jpg --style-image images/9styles/candy.jpg
+
+* ``--content-image``: path to content image.
+* ``--style-image``: path to style image.
+* ``--output-image``: path for saving the output image.
+* ``--content-size``: the content image size to test on.
+* ``--style-size``: the style image size to test on.
+* ``--cuda``: set it to 1 for running on GPU, 0 for CPU.
+
+.. raw:: html
+
+    <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/g1.jpg" width="260px" /> <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/g2.jpg" width="260px" />
+    <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/g3.jpg" width="260px" />
+    <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/g4.jpg" width="260px" />
+    <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/g5.jpg" width="260px" />
+    <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/g6.jpg" width="260px" />
+    <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/g7.jpg" width="260px" />
+    <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/g8.jpg" width="260px" />
+    <img src ="https://raw.githubusercontent.com/zhanghang1989/PyTorch-Style-Transfer/master/images/g9.jpg" width="260px" />
+

docs/source/experiments/texture.rst

+Deep TEN: Deep Texture Encoding Network Example
+===============================================
+
+.. image:: http://hangzh.com/figure/cvpr17.svg
+        :width: 100%
+        :align: left
+
+In this section, we show an example of training/testing Encoding-Net for texture recognition on MINC-2500 dataset. Comparing to original Torch implementation, we use *different learning rate* for pre-trained base network and encoding layer (10x), disable color jittering after reducing lr and adopt much *smaller training image size* (224 instead of 352). 
+
+
+.. note::
+    **Make Sure** to `Install PyTorch Encoding <../notes/compile.html>`_ First.
+
+Test Pre-trained Model
+----------------------
+
+
+- Clone the GitHub repo (I am sure you did during the installation)::
+
+    git clone git@github.com:zhanghang1989/PyTorch-Encoding.git
+
+- Download the `MINC-2500 <http://opensurfaces.cs.cornell.edu/publications/minc/>`_ dataset to ``$HOME/data/minc`` folder. Download pre-trained model (training `curve`_ as bellow, pre-trained on train-1 split using single training size of 224, with an error rate of :math:`19.98\%` using single crop on test-1 set)::
+
+    cd PyTorch-Encoding/experiments
+    bash model/download_models.sh
+
+.. _curve:
+
+.. image:: ../_static/img/deep_ten_curve.svg
+    :width: 70%
+
+- Test pre-trained model on MINC-2500::
+
+    >>> python main.py --dataset minc --model encodingnet --resume model/minc.pth.tar --eval
+    # Teriminal Output:
+    #[======================================== 23/23 ===================================>...]    Step: 104ms | Tot: 3s256ms | Loss: 0.719 | Err: 19.983% (1149/5750) 
+
+
+Train Your Own Model
+--------------------
+
+- Example training command::
+
+    python main.py --dataset minc --model encodingnet --batch-size 64 --lr 0.01 --epochs 60 
+
+- Training options::
+
+  -h, --help            show this help message and exit
+  --dataset DATASET     training dataset (default: cifar10)
+  --model MODEL         network model type (default: densenet)
+  --widen N             widen factor of the network (default: 4)
+  --batch-size N        batch size for training (default: 128)
+  --test-batch-size N   batch size for testing (default: 1000)
+  --epochs N            number of epochs to train (default: 300)
+  --start_epoch N       the epoch number to start (default: 0)
+  --lr LR               learning rate (default: 0.1)
+  --momentum M          SGD momentum (default: 0.9)
+  --weight-decay M      SGD weight decay (default: 1e-4)
+  --no-cuda             disables CUDA training
+  --plot                matplotlib
+  --seed S              random seed (default: 1)
+  --resume RESUME       put the path to resuming file if needed
+  --checkname           set the checkpoint name
+  --eval                evaluating
+
+.. todo::
+    Provide example code for extracting features.
+
+Extending the Software
+----------------------
+
+This code includes an integrated pipeline and some visualization tools (progress bar, real-time training curve plots). It is easy to use and extend for your own model or dataset:
+
+- Write your own Dataloader ``mydataset.py`` to ``dataset/`` folder
+
+- Write your own Model ``mymodel.py`` to ``model/`` folder
+
+- Run the program::
+
+    python main.py --dataset mydataset --model mymodel
+
+
+Citation
+--------
+
+.. note::
+    * Hang Zhang, Jia Xue, and Kristin Dana. "Deep TEN: Texture Encoding Network." *The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2017*::
+
+        @InProceedings{Zhang_2017_CVPR,
+        author = {Zhang, Hang and Xue, Jia and Dana, Kristin},
+        title = {Deep TEN: Texture Encoding Network},
+        booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
+        month = {July},
+        year = {2017}
+        }

docs/source/functions.rst

+.. role:: hidden
+    :class: hidden-section
+
+Other Functions
+===============
+
+.. automodule:: encoding.functions
+
+.. currentmodule:: encoding.functions
+
+
+:hidden:`dilatedavgpool2d`
+~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autofunction:: dilatedavgpool2d
+

docs/source/index.rst

-.. Encoding documentation master file, created by
-   sphinx-quickstart on Fri Dec 23 13:31:47 2016.
-   You can adapt this file completely to your liking, but it should at least
-   contain the root `toctree` directive.
+.. Encoding documentation master file
 
 :github_url: https://github.com/zhanghang1989/PyTorch-Encoding
 
-Encoding documentation
+Encoding Documentation
 ======================
 
-PyTorch-Encoding is an optimized PyTorch package using GPU, including Encoding Layer, Multi-GPU Synchronized Batch Normalization.
+Created by `Hang Zhang <http://hangzh.com/>`_
+
+PyTorch-Encoding is an optimized PyTorch package with CUDA backend, including Encoding Layer, Multi-GPU Synchronized Batch Normalization and useful util functions. Example systems are also provided in `experiments section <experiments/texture.html>`_. We hope this software will accelerate your research, please cite our `papers <notes/compile.html>`_. 
 
 .. toctree::
    :glob:
@@ -18,13 +17,23 @@ PyTorch-Encoding is an optimized PyTorch package using GPU, including Encoding L
    notes/*
 
 .. toctree::
-   :maxdepth: 3
+   :maxdepth: 1
    :caption: Package Reference
 
    encoding
    syncbn
    parallel
+   dilated
+   nn
+   functions
+   utils
+
+.. toctree::
+   :glob:
+   :maxdepth: 1
+   :caption: Experiment Systems
 
+   experiments/*
 
 Indices and tables
 ==================

docs/source/nn.rst

+.. role:: hidden
+    :class: hidden-section
+
+Other NN Layers
+===============
+
+.. automodule:: encoding.nn
+
+Customized Layers
+-----------------
+
+:hidden:`Normalize`
+~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: Normalize
+    :members:
+
+:hidden:`View`
+~~~~~~~~~~~~~~
+
+.. autoclass:: View
+    :members:
+
+Standard Layers
+---------------
+
+Standard Layers as in PyTorch but in :class:`encoding.parallel.SelfDataParallel` mode. Use together with SyncBN.
+
+:hidden:`Conv1d`
+~~~~~~~~~~~~~~~~
+
+.. autoclass:: Conv1d
+    :members:
+
+:hidden:`Conv2d`
+~~~~~~~~~~~~~~~~
+
+.. autoclass:: Conv2d
+    :members:
+
+:hidden:`ConvTranspose2d`
+~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: ConvTranspose2d
+    :members:
+
+:hidden:`ReLU`
+~~~~~~~~~~~~~~
+
+.. autoclass:: ReLU
+    :members:
+
+:hidden:`Sigmoid`
+~~~~~~~~~~~~~~~~~
+
+.. autoclass:: Sigmoid
+    :members:
+
+:hidden:`MaxPool2d`
+~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: MaxPool2d
+    :members:
+
+:hidden:`AvgPool2d`
+~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: AvgPool2d
+    :members:
+
+:hidden:`AdaptiveAvgPool2d`
+~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: AdaptiveAvgPool2d
+    :members:
+
+:hidden:`Dropout2d`
+~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: Dropout2d
+    :members:
+
+:hidden:`Linear`
+~~~~~~~~~~~~~~~~
+
+.. autoclass:: Linear
+    :members:
+
+

docs/source/notes/compile.rst

@@ -16,17 +16,21 @@ Install PyTorch-Encoding
 
         python setup.py install
 
-    * On MAC OSX::
+    * On Mac OSX::
 
          MACOSX_DEPLOYMENT_TARGET=10.9 CC=clang CXX=clang++ python setup.py install
 
 - Reference:
-    Hang Zhang, Jia Xue, and Kristin Dana. "Deep TEN: Texture Encoding Network." *The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2017*::
-
-        @InProceedings{Zhang_2017_CVPR,
-        author = {Zhang, Hang and Xue, Jia and Dana, Kristin},
-        title = {Deep TEN: Texture Encoding Network},
-        booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
-        month = {July},
-        year = {2017}
-        }
+
+    .. note::
+        If using the code in your research, please cite our paper.
+
+        * Hang Zhang, Jia Xue, and Kristin Dana. "Deep TEN: Texture Encoding Network." *The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2017*::
+
+            @InProceedings{Zhang_2017_CVPR,
+            author = {Zhang, Hang and Xue, Jia and Dana, Kristin},
+            title = {Deep TEN: Texture Encoding Network},
+            booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
+            month = {July},
+            year = {2017}
+            }

docs/source/notes/syncbn.rst

+Implementing Synchronized Multi-GPU Batch Normalization
+=======================================================
+We will release the implementation detail of Multi-GPU Batch Normalization in later version.
+
+Why Synchronize?
+----------------
+
+- Standard Implementation
+
+How to Synchronize?
+-------------------
+
+- Forward and Backward Pass
+
+- Synchronized DataParallel
+
+- Cross GPU Autograd
+
+Comparing Performance 
+---------------------
+

docs/source/parallel.rst

@@ -7,10 +7,8 @@ Data Parallel
 Current PyTorch DataParallel Table is not supporting mutl-gpu loss calculation, which makes the gpu memory usage very in-efficient. We address this issue here by doing CriterionDataParallel. 
 The DataParallel compatible with SyncBN will be released later.
 
-Modules
--------
-
-.. currentmodule:: encoding
+.. automodule:: encoding.parallel
+.. currentmodule:: encoding.parallel
 
 :hidden:`ModelDataParallel`
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -24,3 +22,21 @@ Modules
 .. autoclass:: CriterionDataParallel
     :members:
 
+:hidden:`SelfDataParallel`
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: SelfDataParallel
+    :members:
+
+:hidden:`AllReduce`
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: AllReduce
+    :members:
+
+:hidden:`Broadcast`
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: Broadcast
+    :members:
+

docs/source/syncbn.rst

@@ -8,26 +8,46 @@ The current BN is implementated insynchronized accross the gpus, which is a big
 To synchronize the batchnorm accross multiple gpus is not easy to implment within the current Dataparallel framework. We address this difficulty by making each layer 'self-parallel', that is accepting the inputs from multi-gpus. Therefore, we can handle different layers seperately for synchronizing it across gpus.
 We will release the whole SyncBN Module and compatible DataParallel later. 
 
+
+.. currentmodule:: encoding.nn
+
+Modules
+-------
+
+:hidden:`BatchNorm1d`
+~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: BatchNorm1d
+    :members:
+
+:hidden:`BatchNorm2d`
+~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: BatchNorm2d
+    :members:
+
+
 .. currentmodule:: encoding
 
+
 Functions
 ---------
 
+.. currentmodule:: encoding.functions
+
+
 :hidden:`batchnormtrain`
 ~~~~~~~~~~~~~~~~~~~~~~~~
 
-.. autoclass:: batchnormtrain
-    :members:
+.. autofunction:: batchnormtrain
 
 :hidden:`batchnormeval`
 ~~~~~~~~~~~~~~~~~~~~~~~
 
-.. autoclass:: batchnormeval
-    :members:
+.. autofunction:: batchnormeval
 
 :hidden:`sum_square`
 ~~~~~~~~~~~~~~~~~~~~
 
-.. autoclass:: sum_square
-    :members:
+.. autofunction:: sum_square
 

docs/source/utils.rst

+.. role:: hidden
+    :class: hidden-section
+
+My PyTorch Utils
+================
+
+Useful util functions.
+
+.. automodule:: encoding.utils
+.. currentmodule:: encoding.utils
+
+:hidden:`CosLR_Scheduler`
+~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: CosLR_Scheduler
+    :members:
+
+:hidden:`get_optimizer`
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autofunction:: get_optimizer
+
+:hidden:`save_checkpoint`
+~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autofunction:: save_checkpoint
+
+:hidden:`progress_bar`
+~~~~~~~~~~~~~~~~~~~~~~
+
+.. autofunction:: progress_bar

encoding/CMakeLists.txt

@@ -75,6 +75,8 @@ IF(ENCODING_SO_VERSION)
     SOVERSION ${ENCODING_SO_VERSION})
 ENDIF(ENCODING_SO_VERSION)
 
+FILE(GLOB src-header kernel/generic/*.h)
+
 INSTALL(TARGETS ENCODING LIBRARY DESTINATION ${ENCODING_INSTALL_LIB_SUBDIR})
 INSTALL(FILES kernel/thc_encoding.h DESTINATION "${ENCODING_INSTALL_INCLUDE_SUBDIR}/ENCODING")
-INSTALL(FILES kernel/generic/encoding_kernel.h DESTINATION "${ENCODING_INSTALL_INCLUDE_SUBDIR}/ENCODING/generic")
+INSTALL(FILES ${src-header} DESTINATION "${ENCODING_INSTALL_INCLUDE_SUBDIR}/ENCODING/generic")

encoding/__init__.py

@@ -8,8 +8,10 @@
 ## LICENSE file in the root directory of this source tree 
 ##+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 
-from .functions import *
-from .modules import *
-from .syncbn import sum_square, batchnormtrain, batchnormeval
-from .parallel import ModelDataParallel, CriterionDataParallel
+__version__ = '1.0.1'
+
+import encoding.nn
+import encoding.functions
+import encoding.dilated
+import encoding.parallel