Re-write examples (#496)

* Initial commit * Update example tests * Update example tests * Small fix * Small fix * Small fix * Fixes * Fixes * Try something * Try something * Try something * Try something * Try something * Try something * Some docs refactor * Updates * Add tabular * refactor and add text classification * Add multi label text * Updates * Add summarization * Update translation * Add video * Updates * Updates * Update API references * Fix some links * Merge image and video * Merge image and video
Lightning-Universe · Jul 7, 2021 · 5d98cde · 5d98cde
1 parent fb08447
commit 5d98cde
Show file tree

Hide file tree

Showing 75 changed files with 714 additions and 1,803 deletions.
diff --git a/docs/source/_templates/layout.html b/docs/source/_templates/layout.html
@@ -0,0 +1,10 @@
+{% extends "!layout.html" %}
+<link rel="canonical" href="{{ theme_canonical_url }}{{ pagename }}.html" />
+
+{% block footer %}
+{{ super() }}
+<script script type="text/javascript">
+  var collapsedSections = ['Guides', 'Image and Video', 'Tabular', 'Text', 'Integrations', 'API Reference', 'Contributing a Task'];
+</script>
+
+{% endblock %}
diff --git a/docs/source/_templates/theme_variables.jinja b/docs/source/_templates/theme_variables.jinja
@@ -14,5 +14,7 @@
   'blog': 'https://www.pytorchlightning.ai/blog',
   'resources': 'https://lightning-flash.readthedocs.io/',
   'support': 'https://lightning-flash.readthedocs.io/',
+  'community': 'https://pytorch-lightning.slack.com',
+  'forums': 'https://pytorch-lightning.slack.com',
 }
 -%}
diff --git a/docs/source/common/image_backbones.rst b/docs/source/common/image_backbones.rst
diff --git a/docs/source/common/object_detection_backbones.rst b/docs/source/common/object_detection_backbones.rst
diff --git a/docs/source/index.rst b/docs/source/index.rst
@@ -8,7 +8,7 @@ Lightning Flash
 
 .. toctree::
    :maxdepth: 1
-   :caption: Get started:
+   :caption: Getting started
 
    quickstart
    installation
@@ -18,58 +18,61 @@ Lightning Flash
 
 .. toctree::
    :maxdepth: 1
-   :caption: Tasks
+   :caption: Guides
+
+   general/training
+   general/finetuning
+   general/predictions
+   general/jit
+
+.. toctree::
+   :maxdepth: 1
+   :caption: Image and Video
 
-   reference/task
    reference/image_classification
    reference/image_classification_multi_label
    reference/image_embedder
-   reference/summarization
-   reference/text_classification
-   reference/text_classification_multi_label
-   reference/tabular_classification
-   reference/translation
    reference/object_detection
-   reference/video_classification
    reference/semantic_segmentation
    reference/style_transfer
+   reference/video_classification
 
 .. toctree::
    :maxdepth: 1
-   :caption: Flash API
+   :caption: Tabular
 
-   general/data
-   general/callback
-   general/registry
-   general/serve
+   reference/tabular_classification
 
+.. toctree::
+   :maxdepth: 1
+   :caption: Text
+
+   reference/text_classification
+   reference/text_classification_multi_label
+   reference/summarization
+   reference/translation
 
 .. toctree::
    :maxdepth: 1
-   :caption: Common Use Cases
+   :caption: Integrations
 
-   general/training
-   general/finetuning
-   general/predictions
-   general/jit
+   integrations/fiftyone
 
 .. toctree::
    :maxdepth: 1
    :caption: API Reference
 
+   general/data
+   general/callback
+   general/registry
+   general/serve
    code/core
    code/data
    code/image
    code/tabular
    code/text
    code/video
 
-.. toctree::
-   :maxdepth: 1
-   :caption: Integrations
-
-   integrations/fiftyone
-
 .. toctree::
    :maxdepth: 1
    :caption: Contributing a Task

diff --git a/docs/source/quickstart.rst b/docs/source/quickstart.rst
@@ -145,7 +145,6 @@ When you have enough data, you're likely better off training from scratch instea
 A few Built-in Tasks
 ====================
 
-- :doc:`General Task <reference/task>`
 - :doc:`ImageClassification <reference/image_classification>`
 - :doc:`ImageEmbedder <reference/image_embedder>`
 - :doc:`TextClassification <reference/text_classification>`

diff --git a/docs/source/reference/flash_to_production.rst b/docs/source/reference/flash_to_production.rst
@@ -7,15 +7,15 @@ Flash makes it simple to deploy models in production.
 Server Side
 ^^^^^^^^^^^
 
-.. literalinclude:: ../../../flash_examples/serve/segmentic_segmentation/inference_server.py
+.. literalinclude:: ../../../flash_examples/serve/semantic_segmentation/inference_server.py
     :language: python
     :lines: 14-
 
 
 Client Side
 ^^^^^^^^^^^
 
-.. literalinclude:: ../../../flash_examples/serve/segmentic_segmentation/client.py
+.. literalinclude:: ../../../flash_examples/serve/semantic_segmentation/client.py
     :language: python
     :lines: 14-
 

diff --git a/docs/source/reference/image_classification.rst b/docs/source/reference/image_classification.rst
@@ -6,37 +6,22 @@ Image Classification
 ####################
 
 ********
-The task
+The Task
 ********
-The task of identifying what is in an image is called image classification. Typically, Image Classification is used to identify images containing a single object. The task predicts which ‘class’ the image most likely belongs to with a degree of certainty.  A class is a label that describes what is in an image, such as ‘car’, ‘house’, ‘cat’ etc. For example, we can train the image classifier task on images of ants and it will learn to predict the probability that an image contains an ant.
 
-------
-
-*********
-Inference
-*********
-
-The :class:`~flash.image.ImageClassifier` is already pre-trained on `ImageNet <http://www.image-net.org/>`_, a dataset of over 14 million images.
-
-
-Use the :class:`~flash.image.ImageClassifier` pretrained model for inference on any string sequence using :func:`~flash.image.ImageClassifier.predict`:
-
-.. literalinclude:: ../../../flash_examples/predict/image_classification.py
-    :language: python
-    :lines: 14-
-
-For more advanced inference options, see :ref:`predictions`.
+The task of identifying what is in an image is called image classification.
+Typically, Image Classification is used to identify images containing a single object.
+The task predicts which ‘class’ the image most likely belongs to with a degree of certainty.
+A class is a label that describes what is in an image, such as ‘car’, ‘house’, ‘cat’ etc.
 
 ------
 
-**********
-Finetuning
-**********
+*******
+Example
+*******
 
-Lets say you wanted to develope a model that could determine whether an image contains **ants** or **bees**, using the hymenoptera dataset.
-Once we download the data using :func:`~flash.core.data.download_data`, all we need is the train data and validation data folders to create the :class:`~flash.image.ImageClassificationData`.
-
-.. note:: The dataset contains ``train`` and ``validation`` folders, and then each folder contains a **bees** folder, with pictures of bees, and an **ants** folder with images of, you guessed it, ants.
+Let's look at the task of predicting whether images contain Ants or Bees using the hymenoptera dataset.
+The dataset contains ``train`` and ``validation`` folders, and then each folder contains a **bees** folder, with pictures of bees, and an **ants** folder with images of, you guessed it, ants.
 
 .. code-block::
 
@@ -60,36 +45,32 @@ Once we download the data using :func:`~flash.core.data.download_data`, all we n
             ├── 10870992_eebeeb3a12.jpg
             ...
 
+Once we've downloaded the data using :func:`~flash.core.data.download_data`, we create the :class:`~flash.image.classification.data.ImageClassificationData`.
+We select a pre-trained backbone to use for our :class:`~flash.image.classification.model.ImageClassifier` and fine-tune on the hymenoptera data.
+We then use the trained :class:`~flash.image.classification.model.ImageClassifier` for inference.
+Finally, we save the model.
+Here's the full example:
 
-Now all we need is to train our task!
-
-.. literalinclude:: ../../../flash_examples/finetuning/image_classification.py
+.. literalinclude:: ../../../flash_examples/image_classification.py
     :language: python
     :lines: 14-
 
 ------
 
-*********************
-Changing the backbone
-*********************
-By default, we use a `ResNet-18 <https://arxiv.org/abs/1512.03385>`_ for image classification. You can change the model run by the task by passing in a different backbone.
-
-.. testsetup::
-
-    from flash.core.data.utils import download_data
-    from flash.image import ImageClassificationData, ImageClassifier
-
-    download_data("https://pl-flash-data.s3.amazonaws.com/hymenoptera_data.zip", "data/")
+*******
+Serving
+*******
 
-.. testcode::
+The :class:`~flash.image.classification.model.ImageClassifier` is servable.
+This means you can call ``.serve`` to serve your :class:`~flash.Task`.
+Here's an example:
 
-    # 1. organize the data
-    data = ImageClassificationData.from_folders(
-        train_folder="data/hymenoptera_data/train/",
-        val_folder="data/hymenoptera_data/val/",
-    )
+.. literalinclude:: ../../../flash_examples/serve/image_classification/inference_server.py
+    :language: python
+    :lines: 14-
 
-    # 2. build the task
-    task = ImageClassifier(num_classes=2, backbone="resnet34")
+You can now perform inference from your client like this:
 
-.. include:: ../common/image_backbones.rst
+.. literalinclude:: ../../../flash_examples/serve/image_classification/client.py
+    :language: python
+    :lines: 14-
diff --git a/docs/source/reference/image_classification_multi_label.rst b/docs/source/reference/image_classification_multi_label.rst
@@ -6,45 +6,23 @@ Multi-label Image Classification
 ################################
 
 ********
-The task
+The Task
 ********
-Multi-label classification is the task of assigning a number of labels from a fixed set to each data point, which can be in any modality. In this example, we will look at the task of trying to predict the movie genres from an image of the movie poster.
 
-------
-
-********
-The data
-********
-The data we will use in this example is a subset of the awesome movie poster genre prediction data set from the paper "Movie Genre Classification based on Poster Images with Deep Neural Networks" by Wei-Ta Chu and Hung-Jui Guo, resized to 128 by 128.
-Take a look at their paper (and please consider citing their paper if you use the data) here: `www.cs.ccu.edu.tw/~wtchu/projects/MoviePoster/ <https://www.cs.ccu.edu.tw/~wtchu/projects/MoviePoster/>`_.
-
-------
-
-*********
-Inference
-*********
-
-The :class:`~flash.image.ImageClassifier` is already pre-trained on `ImageNet <http://www.image-net.org/>`_, a dataset of over 14 million images.
-
-We can use the :class:`~flash.image.ImageClassifier` model (pretrained on our data) for inference on any string sequence using :func:`~flash.image.ImageClassifier.predict`.
-We can also add a simple visualisation by extending :class:`~flash.core.data.base_viz.BaseVisualization`, like this:
-
-.. literalinclude:: ../../../flash_examples/predict/image_classification_multi_label.py
-    :language: python
-    :lines: 14-
-
-For more advanced inference options, see :ref:`predictions`.
+Multi-label classification is the task of assigning a number of labels from a fixed set to each data point, which can be in any modality (images in this case).
+Multi-label image classification is supported by the :class:`~flash.image.classification.model.ImageClassifier` via the ``multi-label`` argument.
 
 ------
 
-**********
-Finetuning
-**********
+*******
+Example
+*******
 
-Now let's look at how we can finetune a model on the movie poster data.
-Once we download the data using :func:`~flash.core.data.download_data`, all we need is the train data and validation data folders to create the :class:`~flash.image.ImageClassificationData`.
-
-.. note:: The dataset contains ``train`` and ``validation`` folders, and then each folder contains images and a ``metadata.csv`` which stores the labels.
+Let's look at the task of trying to predict the movie genres from an image of the movie poster.
+The data we will use is a subset of the awesome movie poster genre prediction data set from the paper "Movie Genre Classification based on Poster Images with Deep Neural Networks" by Wei-Ta Chu and Hung-Jui Guo, resized to 128 by 128.
+Take a look at their paper (and please consider citing their paper if you use the data) here: `www.cs.ccu.edu.tw/~wtchu/projects/MoviePoster/ <https://www.cs.ccu.edu.tw/~wtchu/projects/MoviePoster/>`_.
+The data set contains ``train`` and ``validation`` folders, and then each folder contains images and a ``metadata.csv`` which stores the labels.
+Here's an overview:
 
 .. code-block::
 
@@ -60,13 +38,22 @@ Once we download the data using :func:`~flash.core.data.download_data`, all we n
         ├── tt0326965.jpg
         ...
 
+Once we've downloaded the data using :func:`~flash.core.data.download_data`, we need to create the :class:`~flash.image.classification.data.ImageClassificationData`.
+We first create a function (``load_data``) to extract the list of images and associated labels which can then be passed to :meth:`~flash.image.classification.data.ImageClassificationData.from_files`.
+We select a pre-trained backbone to use for our :class:`~flash.image.classification.model.ImageClassifier` and fine-tune on the posters data.
+We then use the trained :class:`~flash.image.classification.model.ImageClassifier` for inference.
+Finally, we save the model.
+Here's the full example:
 
-The ``metadata.csv`` files in each folder contain our labels, so we need to create a function (``load_data``) to extract the list of images and associated labels:
-
-.. literalinclude:: ../../../flash_examples/finetuning/image_classification_multi_label.py
+.. literalinclude:: ../../../flash_examples/image_classification_multi_label.py
     :language: python
     :lines: 14-
 
 ------
 
-For more backbone options, see :ref:`image_classification`.
+*******
+Serving
+*******
+
+The :class:`~flash.image.classification.model.ImageClassifier` is servable.
+For more information, see :ref:`image_classification`.