Automated documentation update

PiperOrigin-RevId: 300826885

Author: Conchylicultor

docs/catalog/_toc.yaml

@@ -12,6 +12,8 @@ toc:
     title: ljspeech
   - path: /datasets/catalog/nsynth
     title: nsynth
+  - path: /datasets/catalog/savee
+    title: savee (manual)
   - path: /datasets/catalog/speech_commands
     title: speech_commands
   title: Audio
@@ -98,6 +100,8 @@ toc:
     title: flic
   - path: /datasets/catalog/food101
     title: food101
+  - path: /datasets/catalog/geirhos_conflict_stimuli
+    title: geirhos_conflict_stimuli
   - path: /datasets/catalog/horses_or_humans
     title: horses_or_humans
   - path: /datasets/catalog/i_naturalist2017

docs/catalog/geirhos_conflict_stimuli.md

@@ -0,0 +1,70 @@
+<div itemscope itemtype="http://schema.org/Dataset">
+  <div itemscope itemprop="includedInDataCatalog" itemtype="http://schema.org/DataCatalog">
+    <meta itemprop="name" content="TensorFlow Datasets" />
+  </div>
+
+  <meta itemprop="name" content="geirhos_conflict_stimuli" />
+  <meta itemprop="description" content="Shape/texture conflict stimuli from &quot;ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness.&quot;&#10;&#10;Note that, although the dataset source contains images with matching shape and&#10;texture and we include them here, they are ignored for most evaluations in the&#10;original paper.&#10;&#10;&#10;To use this dataset:&#10;&#10;```python&#10;import tensorflow_datasets as tfds&#10;&#10;ds = tfds.load(&#x27;geirhos_conflict_stimuli&#x27;, split=&#x27;train&#x27;)&#10;for ex in ds.take(4):&#10;  print(ex)&#10;```&#10;&#10;See [the guide](https://www.tensorflow.org/datasets/overview) for more&#10;informations on [tensorflow_datasets](https://www.tensorflow.org/datasets).&#10;&#10;" />
+  <meta itemprop="url" content="https://www.tensorflow.org/datasets/catalog/geirhos_conflict_stimuli" />
+  <meta itemprop="sameAs" content="https://github.com/rgeirhos/texture-vs-shape" />
+  <meta itemprop="citation" content="@inproceedings{&#10;  geirhos2018imagenettrained,&#10;  title={ImageNet-trained {CNN}s are biased towards texture; increasing shape&#10;         bias improves accuracy and robustness.},&#10;  author={Robert Geirhos and Patricia Rubisch and Claudio Michaelis and&#10;          Matthias Bethge and Felix A. Wichmann and Wieland Brendel},&#10;  booktitle={International Conference on Learning Representations},&#10;  year={2019},&#10;  url={https://openreview.net/forum?id=Bygh9j09KX},&#10;}&#10;" />
+</div>
+
+# `geirhos_conflict_stimuli`
+
+*   **Description**:
+
+Shape/texture conflict stimuli from "ImageNet-trained CNNs are biased towards
+texture; increasing shape bias improves accuracy and robustness."
+
+Note that, although the dataset source contains images with matching shape and
+texture and we include them here, they are ignored for most evaluations in the
+original paper.
+
+*   **Homepage**:
+    [https://github.com/rgeirhos/texture-vs-shape](https://github.com/rgeirhos/texture-vs-shape)
+*   **Source code**:
+    [`tfds.image.geirhos_conflict_stimuli.GeirhosConflictStimuli`](https://github.com/tensorflow/datasets/tree/master/tensorflow_datasets/image/geirhos_conflict_stimuli.py)
+*   **Versions**:
+    *   **`1.0.0`** (default): No release notes.
+*   **Download size**: `153.96 MiB`
+*   **Dataset size**: `130.44 MiB`
+*   **Auto-cached**
+    ([documentation](https://www.tensorflow.org/datasets/performances#auto-caching)):
+    Only when `shuffle_files=False` (test)
+*   **Splits**:
+
+Split  | Examples
+:----- | -------:
+'test' | 1,280
+
+*   **Features**:
+
+```python
+FeaturesDict({
+    'file_name': Text(shape=(), dtype=tf.string),
+    'image': Image(shape=(None, None, 3), dtype=tf.uint8),
+    'shape_imagenet_labels': Sequence(ClassLabel(shape=(), dtype=tf.int64, num_classes=1000)),
+    'shape_label': ClassLabel(shape=(), dtype=tf.int64, num_classes=16),
+    'texture_imagenet_labels': Sequence(ClassLabel(shape=(), dtype=tf.int64, num_classes=1000)),
+    'texture_label': ClassLabel(shape=(), dtype=tf.int64, num_classes=16),
+})
+```
+
+*   **Supervised keys** (See
+    [`as_supervised` doc](https://www.tensorflow.org/datasets/api_docs/python/tfds/load#args)):
+    `('image', 'shape_label')`
+*   **Citation**:
+
+```
+@inproceedings{
+  geirhos2018imagenettrained,
+  title={ImageNet-trained {CNN}s are biased towards texture; increasing shape
+         bias improves accuracy and robustness.},
+  author={Robert Geirhos and Patricia Rubisch and Claudio Michaelis and
+          Matthias Bethge and Felix A. Wichmann and Wieland Brendel},
+  booktitle={International Conference on Learning Representations},
+  year={2019},
+  url={https://openreview.net/forum?id=Bygh9j09KX},
+}
+```

docs/catalog/libritts.md

@@ -29,16 +29,23 @@ The main differences from the LibriSpeech corpus are listed below:
 *   **Source code**:
     [`tfds.audio.libritts.Libritts`](https://github.com/tensorflow/datasets/tree/master/tensorflow_datasets/audio/libritts.py)
 *   **Versions**:
-    *   **`1.0.0`** (default): No release notes.
-*   **Download size**: `Unknown size`
-*   **Dataset size**: `Unknown size`
+    *   **`1.0.1`** (default): No release notes.
+*   **Download size**: `78.42 GiB`
+*   **Dataset size**: `271.41 GiB`
 *   **Auto-cached**
     ([documentation](https://www.tensorflow.org/datasets/performances#auto-caching)):
-    Unknown
+    No
 *   **Splits**:
 
-Split | Examples
-:---- | -------:
+Split            | Examples
+:--------------- | -------:
+'dev_clean'      | 5,736
+'dev_other'      | 4,613
+'test_clean'     | 4,837
+'test_other'     | 5,120
+'train_clean100' | 33,236
+'train_clean360' | 116,500
+'train_other500' | 205,044
 
 *   **Features**:
 

docs/catalog/overview.md

@@ -40,6 +40,7 @@ np_datasets = tfds.as_numpy(datasets)
     *   [`libritts`](libritts.md)
     *   [`ljspeech`](ljspeech.md)
     *   [`nsynth`](nsynth.md)
+    *   [`savee`](savee.md)
     *   [`speech_commands`](speech_commands.md)
 *   `Image`
     *   [`abstract_reasoning`](abstract_reasoning.md)
@@ -83,6 +84,7 @@ np_datasets = tfds.as_numpy(datasets)
     *   [`fashion_mnist`](fashion_mnist.md)
     *   [`flic`](flic.md)
     *   [`food101`](food101.md)
+    *   [`geirhos_conflict_stimuli`](geirhos_conflict_stimuli.md)
     *   [`horses_or_humans`](horses_or_humans.md)
     *   [`i_naturalist2017`](i_naturalist2017.md)
     *   [`image_label_folder`](image_label_folder.md)

docs/catalog/robonet.md

@@ -51,15 +51,16 @@ from 113 unique camera viewpoints.
 ## robonet/robonet_sample_64 (default config)
 
 *   **Config description**: 64x64 RoboNet Sample.
-*   **Download size**: `Unknown size`
-*   **Dataset size**: `Unknown size`
+*   **Download size**: `119.80 MiB`
+*   **Dataset size**: `183.01 MiB`
 *   **Auto-cached**
     ([documentation](https://www.tensorflow.org/datasets/performances#auto-caching)):
-    Unknown
+    Only when `shuffle_files=False` (train)
 *   **Splits**:
 
-Split | Examples
-:---- | -------:
+Split   | Examples
+:------ | -------:
+'train' | 700
 
 *   **Features**:
 
@@ -74,15 +75,16 @@ FeaturesDict({
 ## robonet/robonet_sample_128
 
 *   **Config description**: 128x128 RoboNet Sample.
-*   **Download size**: `Unknown size`
-*   **Dataset size**: `Unknown size`
+*   **Download size**: `119.80 MiB`
+*   **Dataset size**: `638.95 MiB`
 *   **Auto-cached**
     ([documentation](https://www.tensorflow.org/datasets/performances#auto-caching)):
-    Unknown
+    No
 *   **Splits**:
 
-Split | Examples
-:---- | -------:
+Split   | Examples
+:------ | -------:
+'train' | 700
 
 *   **Features**:
 
@@ -121,15 +123,16 @@ FeaturesDict({
 ## robonet/robonet_128
 
 *   **Config description**: 128x128 RoboNet.
-*   **Download size**: `Unknown size`
-*   **Dataset size**: `Unknown size`
+*   **Download size**: `36.20 GiB`
+*   **Dataset size**: `144.90 GiB`
 *   **Auto-cached**
     ([documentation](https://www.tensorflow.org/datasets/performances#auto-caching)):
-    Unknown
+    No
 *   **Splits**:
 
-Split | Examples
-:---- | -------:
+Split   | Examples
+:------ | -------:
+'train' | 162,417
 
 *   **Features**:
 

docs/catalog/savee.md

@@ -0,0 +1,75 @@
+<div itemscope itemtype="http://schema.org/Dataset">
+  <div itemscope itemprop="includedInDataCatalog" itemtype="http://schema.org/DataCatalog">
+    <meta itemprop="name" content="TensorFlow Datasets" />
+  </div>
+
+  <meta itemprop="name" content="savee" />
+  <meta itemprop="description" content="&#10;SAVEE (Surrey Audio-Visual Expressed Emotion) is an emotion recognition&#10;dataset. It consists of recordings from 4 male actors in 7 different emotions,&#10;480 British English utterances in total. The sentences were chosen from the&#10;standard TIMIT corpus and phonetically-balanced for each emotion.&#10;This release contains only the audio stream from the original audio-visual&#10;recording.&#10;The data is split so that the training set consists of 2 speakers, and both the&#10;validation and test set consists of samples from 1 speaker, respectively.&#10;&#10;&#10;To use this dataset:&#10;&#10;```python&#10;import tensorflow_datasets as tfds&#10;&#10;ds = tfds.load(&#x27;savee&#x27;, split=&#x27;train&#x27;)&#10;for ex in ds.take(4):&#10;  print(ex)&#10;```&#10;&#10;See [the guide](https://www.tensorflow.org/datasets/overview) for more&#10;informations on [tensorflow_datasets](https://www.tensorflow.org/datasets).&#10;&#10;" />
+  <meta itemprop="url" content="https://www.tensorflow.org/datasets/catalog/savee" />
+  <meta itemprop="sameAs" content="http://kahlan.eps.surrey.ac.uk/savee/" />
+  <meta itemprop="citation" content="&#10;@inproceedings{Vlasenko_combiningframe,&#10;author = {Vlasenko, Bogdan and Schuller, Bjorn and Wendemuth, Andreas and Rigoll, Gerhard},&#10;year = {2007},&#10;month = {01},&#10;pages = {2249-2252},&#10;title = {Combining frame and turn-level information for robust recognition of emotions within speech},&#10;journal = {Proceedings of Interspeech}&#10;}&#10;" />
+</div>
+
+# `savee`
+
+Warning: Manual download required. See instructions below.
+
+*   **Description**:
+
+SAVEE (Surrey Audio-Visual Expressed Emotion) is an emotion recognition dataset.
+It consists of recordings from 4 male actors in 7 different emotions, 480
+British English utterances in total. The sentences were chosen from the standard
+TIMIT corpus and phonetically-balanced for each emotion. This release contains
+only the audio stream from the original audio-visual recording. The data is
+split so that the training set consists of 2 speakers, and both the validation
+and test set consists of samples from 1 speaker, respectively.
+
+*   **Homepage**:
+    [http://kahlan.eps.surrey.ac.uk/savee/](http://kahlan.eps.surrey.ac.uk/savee/)
+*   **Source code**:
+    [`tfds.audio.savee.Savee`](https://github.com/tensorflow/datasets/tree/master/tensorflow_datasets/audio/savee.py)
+*   **Versions**:
+    *   **`1.0.0`** (default): No release notes.
+*   **Download size**: `Unknown size`
+*   **Dataset size**: `Unknown size`
+*   **Manual download instructions**: This dataset requires you to download the
+    source data manually into `download_config.manual_dir`
+    (defaults to `~/tensorflow_datasets/manual/savee/`):<br/>
+    manual_dir should contain the file AudioData.zip. This file should be under
+    Data/Zip/AudioData.zip in the dataset folder provided upon registration.
+    You need to register at
+    http://personal.ee.surrey.ac.uk/Personal/P.Jackson/SAVEE/Register.html in
+    order to get the link to download the dataset.
+*   **Auto-cached**
+    ([documentation](https://www.tensorflow.org/datasets/performances#auto-caching)):
+    Unknown
+*   **Splits**:
+
+Split | Examples
+:---- | -------:
+
+*   **Features**:
+
+```python
+FeaturesDict({
+    'audio': Audio(shape=(None,), dtype=tf.int64),
+    'label': ClassLabel(shape=(), dtype=tf.int64, num_classes=7),
+    'speaker_id': tf.string,
+})
+```
+
+*   **Supervised keys** (See
+    [`as_supervised` doc](https://www.tensorflow.org/datasets/api_docs/python/tfds/load#args)):
+    `('audio', 'label')`
+*   **Citation**:
+
+```
+@inproceedings{Vlasenko_combiningframe,
+author = {Vlasenko, Bogdan and Schuller, Bjorn and Wendemuth, Andreas and Rigoll, Gerhard},
+year = {2007},
+month = {01},
+pages = {2249-2252},
+title = {Combining frame and turn-level information for robust recognition of emotions within speech},
+journal = {Proceedings of Interspeech}
+}
+```