Updating documentation.

docs/conf.py

@@ -36,6 +36,8 @@ extensions = [
     'sphinx.ext.viewcode',
     'sphinx.ext.intersphinx',
     'numpydoc',
+    'IPython.sphinxext.ipython_directive',
+    'IPython.sphinxext.ipython_console_highlighting'
 ]
 
 # numpydoc options

docs/index.rst

@@ -18,7 +18,6 @@ Using PySegCNN
 .. toctree::
    :maxdepth: 1
 
-   ./source/requirements
    ./source/installation
    ./source/api
 

docs/source/api.rst

@@ -5,31 +5,29 @@
 API Reference
 =============
 
-This page lists the available functions and classes of ``pysegcnn``.
-
+This page lists functions and classes of the ``pysegcnn`` package, which are
+relevant at the API level. If you need to dig deeper into the source files, go
+to the git `repository <https://gitlab.inf.unibz.it/REMSEN/ccisnow/pysegcnn>`_.
 
 Dataset
 -------
 
-Custom dataset classes compliant to the PyTorch `standard <https://pytorch.org/docs/stable/data.html>`_.
+The ``pysegcnn`` package offers support for custom datasets, which are compliant
+to the PyTorch `standard <https://pytorch.org/docs/stable/data.html>`_.
+Currently, image datasets from the `Landsat <https://landsat.usgs.gov/>`_ and
+`Sentinel-2`_ satellites are supported.
 
 Generic classes
 ^^^^^^^^^^^^^^^
-Generic class to implement a `custom dataset <https://pytorch.org/tutorials/beginner/data_loading_tutorial.html>`_.
-
-.. autosummary::
-    :toctree: generated/
-    :nosignatures:
-
-    core.dataset.ImageDataset
 
-Generic class to implement a custom dataset following a standard directory
-structure.
+The following two generic classes can be inherited to implement a
+`custom dataset <https://pytorch.org/tutorials/beginner/data_loading_tutorial.html>`_.
 
 .. autosummary::
     :toctree: generated/
     :nosignatures:
 
+    core.dataset.ImageDataset
     core.dataset.StandardEoDataset
 
 
@@ -49,38 +47,24 @@ spaceborne multispectral image datasets are supported out-of-the-box:
 Models
 ------
 
-Layers
-^^^^^^
-
-Convolutional neural network layers.
+The ``pysegcnn`` package ships with a customizable interface to build
+convolutional neural networks for image segmentation.
 
 .. autosummary::
     :toctree: generated/
     :nosignatures:
 
-    core.layers.Block
     core.layers.Conv2dSame
+    core.layers.Block
     core.layers.ConvBnReluMaxPool
     core.layers.ConvBnReluMaxUnpool
     core.layers.ConvBnReluUpsample
-
-Encoder-Decoder architechture
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
-
-Generic ``Encoder`` and ``Decoder`` classes to build an encoder-decoder
-architecture.
-
-.. autosummary::
-    :toctree: generated/
-    :nosignatures:
-
     core.layers.EncoderBlock
     core.layers.DecoderBlock
     core.layers.Encoder
     core.layers.Decoder
 
-Neural Networks
-^^^^^^^^^^^^^^^
+Implemented models.
 
 .. autosummary::
     :toctree: generated/
@@ -90,6 +74,19 @@ Neural Networks
     core.models.UNet
 
 
+Training, Validation and Test set
+---------------------------------
+
+Classes to split a dataset into training, validation and test set.
+
+.. autosummary::
+    :toctree: generated/
+    :nosignatures:
+
+    core.split.DateSplit
+    core.split.RandomTileSplit
+    core.split.RandomSceneSplit
+
 ..
     Links:
 
@@ -113,3 +110,6 @@ Neural Networks
 
 .. _Hughes & Hayes (2014):
     https://www.mdpi.com/2072-4292/6/6/4907
+
+.. _Early Stopping:
+    https://en.wikipedia.org/wiki/Early_stopping

docs/source/installation.rst

@@ -3,6 +3,22 @@
 Installation
 ============
 
+Requirements
+------------
+``pysegcnn`` is a pure Python package that works on both Windows and Linux.
+
+.. important::
+
+    ``pysegcnn`` requires ``Python>=3.7``.
+
+Here is a list of all dependencies of ``pysegcnn``.
+
+    - numpy
+    - scipy
+    - matplotlib
+    - pytorch
+    - gdal
+
 Download
 ---------
 You can download ``pysegcnn`` from this repository's
@@ -74,6 +90,6 @@ root directory within the activated ``pysegcnn`` conda environment**:
 If successful, you should be able to import ``pysegcnn`` from any Python
 interpreter using:
 
-.. code-block:: python
+.. ipython:: python
 
     import pysegcnn

docs/source/requirements.rst

-Requirements
-============
-``pysegcnn`` is a pure Python package that works on both Windows and Linux.
-
-.. important:
-
-    ``pysegcnn`` requires ``Python >= 3.7``.
-
-Dependencies
-^^^^^^^^^^^^
-
-Here is a list of all dependencies of ``pysegcnn``.
-
-Machine learning framework
-^^^^^^^^^^^^^^^^^^^^^^^^^^
-    - pytorch
-
-SciPy stack
-^^^^^^^^^^^
-    - numpy
-    - scipy
-    - matplotlib
-
-Raster tools
-^^^^^^^^^^^^
-    - gdal
-
-Testing
-^^^^^^^
-    - pytest
-
-And if you want to build the documentation:
-    - sphinx
-    - sphinx_rtd_theme
-
-If you are using `conda <https://docs.conda.io/en/latest/miniconda.html>`_,
-you can use the ``environment.yml`` file that ships with ``pysegcnn`` to take
-care of the dependencies, see :doc:`installation`.
\ No newline at end of file