[Docs] Fix docs.

docs/en/conf.py

@@ -129,7 +129,7 @@ html_theme_options = {
 # so a file named "default.css" will overwrite the builtin "default.css".
 html_static_path = ['_static']
 html_css_files = ['css/readthedocs.css']
-#  html_js_files = ['js/custom.js']
+html_js_files = ['js/custom.js']
 
 # -- Options for HTMLHelp output ---------------------------------------------
 
@@ -236,5 +236,4 @@ def builder_inited_handler(app):
 
 
 def setup(app):
-    app.add_js_file('./_static/js/custom.js')
     app.connect('builder-inited', builder_inited_handler)

docs/zh_CN/_static/css/readthedocs.css

@@ -14,3 +14,7 @@ article.pytorch-article .section :not(dt) > code {
   background-color: #f3f4f7;
   border-radius: 5px;
 }
+
+table.colwidths-auto td {
+  width: 50%
+}

docs/zh_CN/_static/js/custom.js

@@ -0,0 +1 @@
+var collapsedSections = ['Model zoo'];

docs/zh_CN/api.rst

@@ -1,68 +0,0 @@
-mmcls.apis
--------------
-.. automodule:: mmcls.apis
-    :members:
-
-mmcls.core
--------------
-
-evaluation
-^^^^^^^^^^
-.. automodule:: mmcls.core.evaluation
-    :members:
-
-mmcls.models
----------------
-
-models
-^^^^^^
-.. automodule:: mmcls.models
-    :members:
-
-classifiers
-^^^^^^^^^^^
-.. automodule:: mmcls.models.classifiers
-    :members:
-
-backbones
-^^^^^^^^^^
-.. automodule:: mmcls.models.backbones
-    :members:
-
-heads
-^^^^^^^^^^
-.. automodule:: mmcls.models.heads
-    :members:
-
-necks
-^^^^^^
-.. automodule:: mmcls.models.necks
-    :members:
-
-losses
-^^^^^^
-.. automodule:: mmcls.models.losses
-    :members:
-
-utils
-^^^^^^
-.. automodule:: mmcls.models.utils
-    :members:
-
-mmcls.datasets
------------------
-
-datasets
-^^^^^^^^
-.. automodule:: mmcls.datasets
-    :members:
-
-pipelines
-^^^^^^^^^
-.. automodule:: mmcls.datasets.pipelines
-    :members:
-
-mmcls.utils
---------------
-.. automodule:: mmcls.utils
-    :members:

docs/zh_CN/api/apis.rst

@@ -0,0 +1,45 @@
+.. role:: hidden
+    :class: hidden-section
+
+mmcls.apis
+===================================
+
+These are some high-level APIs for classification tasks.
+
+.. contents:: mmcls.apis
+   :depth: 2
+   :local:
+   :backlinks: top
+
+.. currentmodule:: mmcls.apis
+
+Train
+------------------
+
+.. autosummary::
+   :toctree: generated
+   :nosignatures:
+
+   init_random_seed
+   set_random_seed
+   train_model
+
+Test
+------------------
+.. autosummary::
+   :toctree: generated
+   :nosignatures:
+
+   single_gpu_test
+   multi_gpu_test
+
+Inference
+------------------
+
+.. autosummary::
+   :toctree: generated
+   :nosignatures:
+
+   init_model
+   inference_model
+   show_result_pyplot

docs/zh_CN/api/core.rst

@@ -0,0 +1,61 @@
+.. role:: hidden
+    :class: hidden-section
+
+mmcls.core
+===================================
+
+This package includes some runtime components. These components are useful in
+classification tasks but not supported by MMCV yet.
+
+.. note::
+
+   Some components may be moved to MMCV in the future.
+
+.. contents:: mmcls.core
+   :depth: 2
+   :local:
+   :backlinks: top
+
+.. currentmodule:: mmcls.core
+
+Evaluation
+------------------
+
+Evaluation metrics calculation functions
+
+.. autosummary::
+   :toctree: generated
+   :nosignatures:
+
+   precision
+   recall
+   f1_score
+   precision_recall_f1
+   average_precision
+   mAP
+   support
+   average_performance
+   calculate_confusion_matrix
+
+Hook
+------------------
+
+.. autosummary::
+   :toctree: generated
+   :nosignatures:
+   :template: classtemplate.rst
+
+   ClassNumCheckHook
+   PreciseBNHook
+   CosineAnnealingCooldownLrUpdaterHook
+
+
+Optimizers
+------------------
+
+.. autosummary::
+   :toctree: generated
+   :nosignatures:
+   :template: classtemplate.rst
+
+   Lamb

docs/zh_CN/api/datasets.rst

@@ -0,0 +1,56 @@
+.. role:: hidden
+    :class: hidden-section
+
+mmcls.datasets
+===================================
+
+The ``datasets`` package contains several usual datasets for image classification tasks and some dataset wrappers.
+
+.. currentmodule:: mmcls.datasets
+
+Custom Dataset
+--------------
+
+.. autoclass:: CustomDataset
+
+ImageNet
+--------
+
+.. autoclass:: ImageNet
+
+.. autoclass:: ImageNet21k
+
+CIFAR
+-----
+
+.. autoclass:: CIFAR10
+
+.. autoclass:: CIFAR100
+
+MNIST
+-----
+
+.. autoclass:: MNIST
+
+.. autoclass:: FashionMNIST
+
+VOC
+---
+
+.. autoclass:: VOC
+
+Base classes
+------------
+
+.. autoclass:: BaseDataset
+
+.. autoclass:: MultiLabelDataset
+
+Dataset Wrappers
+----------------
+
+.. autoclass:: ConcatDataset
+
+.. autoclass:: RepeatDataset
+
+.. autoclass:: ClassBalancedDataset

docs/zh_CN/api/models.rst

@@ -0,0 +1,137 @@
+.. role:: hidden
+    :class: hidden-section
+
+mmcls.models
+===================================
+
+The ``models`` package contains several sub-packages for addressing the different components of a model.
+
+- :ref:`classifiers`: The top-level module which defines the whole process of a classification model.
+- :ref:`backbones`: Usually a feature extraction network, e.g., ResNet, MobileNet.
+- :ref:`necks`: The component between backbones and heads, e.g., GlobalAveragePooling.
+- :ref:`heads`: The component for specific tasks. In MMClassification, we provides heads for classification.
+- :ref:`losses`: Loss functions.
+
+.. currentmodule:: mmcls.models
+
+.. autosummary::
+    :toctree: generated
+    :nosignatures:
+
+    build_classifier
+    build_backbone
+    build_neck
+    build_head
+    build_loss
+
+.. _classifiers:
+
+Classifier
+------------------
+
+.. autosummary::
+   :toctree: generated
+   :nosignatures:
+   :template: classtemplate.rst
+
+   BaseClassifier
+   ImageClassifier
+
+.. _backbones:
+
+Backbones
+------------------
+
+.. autosummary::
+   :toctree: generated
+   :nosignatures:
+   :template: classtemplate.rst
+
+   AlexNet
+   CSPDarkNet
+   CSPNet
+   CSPResNeXt
+   CSPResNet
+   Conformer
+   ConvMixer
+   ConvNeXt
+   DistilledVisionTransformer
+   EfficientNet
+   HRNet
+   LeNet5
+   MlpMixer
+   MobileNetV2
+   MobileNetV3
+   PCPVT
+   PoolFormer
+   RegNet
+   RepMLPNet
+   RepVGG
+   Res2Net
+   ResNeSt
+   ResNeXt
+   ResNet
+   ResNetV1c
+   ResNetV1d
+   ResNet_CIFAR
+   SEResNeXt
+   SEResNet
+   SVT
+   ShuffleNetV1
+   ShuffleNetV2
+   SwinTransformer
+   T2T_ViT
+   TIMMBackbone
+   TNT
+   VGG
+   VisionTransformer
+
+.. _necks:
+
+Necks
+------------------
+
+.. autosummary::
+   :toctree: generated
+   :nosignatures:
+   :template: classtemplate.rst
+
+   GlobalAveragePooling
+   GeneralizedMeanPooling
+   HRFuseScales
+
+.. _heads:
+
+Heads
+------------------
+
+.. autosummary::
+   :toctree: generated
+   :nosignatures:
+   :template: classtemplate.rst
+
+   ClsHead
+   LinearClsHead
+   StackedLinearClsHead
+   MultiLabelClsHead
+   MultiLabelLinearClsHead
+   VisionTransformerClsHead
+   DeiTClsHead
+   ConformerHead
+
+.. _losses:
+
+Losses
+------------------
+
+.. autosummary::
+   :toctree: generated
+   :nosignatures:
+   :template: classtemplate.rst
+
+   Accuracy
+   AsymmetricLoss
+   CrossEntropyLoss
+   LabelSmoothLoss
+   FocalLoss
+   SeesawLoss

docs/zh_CN/api/models.utils.augment.rst

@@ -0,0 +1,35 @@
+.. role:: hidden
+    :class: hidden-section
+
+Batch Augmentation
+===================================
+
+Batch augmentation is the augmentation which involve multiple samples, such as Mixup and CutMix.
+
+In MMClassification, these batch augmentation is used as a part of :ref:`classifiers`. A typical usage is as below:
+
+.. code-block:: python
+
+   model = dict(
+       backbone = ...,
+       neck = ...,
+       head = ...,
+       train_cfg=dict(augments=[
+           dict(type='BatchMixup', alpha=0.8, prob=0.5, num_classes=num_classes),
+           dict(type='BatchCutMix', alpha=1.0, prob=0.5, num_classes=num_classes),
+       ]))
+   )
+
+.. currentmodule:: mmcls.models.utils.augment
+
+Mixup
+-----
+.. autoclass:: BatchMixupLayer
+
+CutMix
+------
+.. autoclass:: BatchCutMixLayer
+
+ResizeMix
+---------
+.. autoclass:: BatchResizeMixLayer

docs/zh_CN/api/models.utils.rst

@@ -0,0 +1,50 @@
+.. role:: hidden
+    :class: hidden-section
+
+mmcls.models.utils
+===================================
+
+This package includes some helper functions and common components used in various networks.
+
+.. contents:: mmcls.models.utils
+   :depth: 2
+   :local:
+   :backlinks: top
+
+.. currentmodule:: mmcls.models.utils
+
+Common Components
+------------------
+
+.. autosummary::
+   :toctree: generated
+   :nosignatures:
+   :template: classtemplate.rst
+
+   InvertedResidual
+   SELayer
+   ShiftWindowMSA
+   MultiheadAttention
+   ConditionalPositionEncoding
+
+Helper Functions
+------------------
+
+channel_shuffle
+^^^^^^^^^^^^^^^
+.. autofunction:: channel_shuffle
+
+make_divisible
+^^^^^^^^^^^^^^
+.. autofunction:: make_divisible
+
+to_ntuple
+^^^^^^^^^^^^^^
+.. autofunction:: to_ntuple
+.. autofunction:: to_2tuple
+.. autofunction:: to_3tuple
+.. autofunction:: to_4tuple
+
+is_tracing
+^^^^^^^^^^^^^^
+.. autofunction:: is_tracing

docs/zh_CN/api/transforms.rst

@@ -0,0 +1,171 @@
+.. role:: hidden
+    :class: hidden-section
+
+Data Transformations
+***********************************
+
+In MMClassification, the data preparation and the dataset is decomposed. The
+datasets only define how to get samples' basic information from the file
+system. These basic information includes the ground-truth label and raw images
+data / the paths of images.
+
+To prepare the inputs data, we need to do some transformations on these basic
+information. These transformations includes loading, preprocessing and
+formatting. And a series of data transformations makes up a data pipeline.
+Therefore, you can find the a ``pipeline`` argument in the configs of dataset,
+for example:
+
+.. code:: python
+
+    img_norm_cfg = dict(
+        mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
+    train_pipeline = [
+        dict(type='LoadImageFromFile'),
+        dict(type='RandomResizedCrop', size=224),
+        dict(type='RandomFlip', flip_prob=0.5, direction='horizontal'),
+        dict(type='Normalize', **img_norm_cfg),
+        dict(type='ImageToTensor', keys=['img']),
+        dict(type='ToTensor', keys=['gt_label']),
+        dict(type='Collect', keys=['img', 'gt_label'])
+    ]
+    test_pipeline = [
+        dict(type='LoadImageFromFile'),
+        dict(type='Resize', size=256),
+        dict(type='CenterCrop', crop_size=224),
+        dict(type='Normalize', **img_norm_cfg),
+        dict(type='ImageToTensor', keys=['img']),
+        dict(type='Collect', keys=['img'])
+    ]
+
+    data = dict(
+        train=dict(..., pipeline=train_pipeline),
+        val=dict(..., pipeline=test_pipeline),
+        test=dict(..., pipeline=test_pipeline),
+    )
+
+Every item of a pipeline list is one of the following data transformations class. And if you want to add a custom data transformation class, the tutorial :doc:`Custom Data Pipelines </tutorials/data_pipeline>` will help you.
+
+.. contents:: mmcls.datasets.pipelines
+   :depth: 2
+   :local:
+   :backlinks: top
+
+.. currentmodule:: mmcls.datasets.pipelines
+
+Loading
+=======
+
+LoadImageFromFile
+---------------------
+.. autoclass:: LoadImageFromFile
+
+Preprocessing and Augmentation
+==============================
+
+CenterCrop
+---------------------
+.. autoclass:: CenterCrop
+
+Lighting
+---------------------
+.. autoclass:: Lighting
+
+Normalize
+---------------------
+.. autoclass:: Normalize
+
+Pad
+---------------------
+.. autoclass:: Pad
+
+Resize
+---------------------
+.. autoclass:: Resize
+
+RandomCrop
+---------------------
+.. autoclass:: RandomCrop
+
+RandomErasing
+---------------------
+.. autoclass:: RandomErasing
+
+RandomFlip
+---------------------
+.. autoclass:: RandomFlip
+
+RandomGrayscale
+---------------------
+.. autoclass:: RandomGrayscale
+
+RandomResizedCrop
+---------------------
+.. autoclass:: RandomResizedCrop
+
+ColorJitter
+---------------------
+.. autoclass:: ColorJitter
+
+
+Composed Augmentation
+---------------------
+Composed augmentation is a kind of methods which compose a series of data
+augmentation transformations, such as ``AutoAugment`` and ``RandAugment``.
+
+.. autoclass:: AutoAugment
+
+.. autoclass:: RandAugment
+
+In composed augmentation, we need to specify several data transformations or
+several groups of data transformations (The ``policies`` argument) as the
+random sampling space. These data transformations are chosen from the below
+table. In addition, we provide some preset policies in `this folder`_.
+
+.. _this folder: https://github.com/open-mmlab/mmclassification/tree/master/configs/_base_/datasets/pipelines
+
+.. autosummary::
+   :toctree: generated
+   :nosignatures:
+   :template: classtemplate.rst
+
+   AutoContrast
+   Brightness
+   ColorTransform
+   Contrast
+   Cutout
+   Equalize
+   Invert
+   Posterize
+   Rotate
+   Sharpness
+   Shear
+   Solarize
+   SolarizeAdd
+   Translate
+
+Formatting
+==========
+
+Collect
+---------------------
+.. autoclass:: Collect
+
+ImageToTensor
+---------------------
+.. autoclass:: ImageToTensor
+
+ToNumpy
+---------------------
+.. autoclass:: ToNumpy
+
+ToPIL
+---------------------
+.. autoclass:: ToPIL
+
+ToTensor
+---------------------
+.. autoclass:: ToTensor
+
+Transpose
+---------------------
+.. autoclass:: Transpose

docs/zh_CN/api/utils.rst

@@ -0,0 +1,23 @@
+.. role:: hidden
+    :class: hidden-section
+
+mmcls.utils
+===================================
+
+These are some useful help function in the ``utils`` package.
+
+.. contents:: mmcls.utils
+   :depth: 1
+   :local:
+   :backlinks: top
+
+.. currentmodule:: mmcls.utils
+
+.. autosummary::
+   :toctree: generated
+   :nosignatures:
+
+   collect_env
+   get_root_logger
+   load_json_log
+   setup_multi_processes

docs/zh_CN/conf.py

@@ -44,6 +44,8 @@ release = get_version()
 # ones.
 extensions = [
     'sphinx.ext.autodoc',
+    'sphinx.ext.autosummary',
+    'sphinx.ext.intersphinx',
     'sphinx.ext.napoleon',
     'sphinx.ext.viewcode',
     'sphinx_markdown_tables',
@@ -127,6 +129,7 @@ html_theme_options = {
 # so a file named "default.css" will overwrite the builtin "default.css".
 html_static_path = ['_static']
 html_css_files = ['css/readthedocs.css']
+html_js_files = ['js/custom.js']
 
 # -- Options for HTMLHelp output ---------------------------------------------
 
@@ -204,6 +207,15 @@ StandaloneHTMLBuilder.supported_image_types = [
 # Ignore >>> when copying code
 copybutton_prompt_text = r'>>> |\.\.\. '
 copybutton_prompt_is_regexp = True
+# Auto-generated header anchors
+myst_heading_anchors = 3
+# Configuration for intersphinx
+intersphinx_mapping = {
+    'python': ('https://docs.python.org/3', None),
+    'numpy': ('https://numpy.org/doc/stable', None),
+    'torch': ('https://pytorch.org/docs/stable/', None),
+    'mmcv': ('https://mmcv.readthedocs.io/en/master/', None),
+}
 
 
 def builder_inited_handler(app):

docs/zh_CN/docutils.conf

@@ -0,0 +1,2 @@
+[html writers]
+table_style: colwidths-auto

docs/zh_CN/index.rst

@@ -59,7 +59,14 @@ You can switch between Chinese and English documentation in the lower-left corne
 .. toctree::
    :caption: API 参考文档
 
-   api.rst
+   mmcls.apis <api/apis>
+   mmcls.core <api/core>
+   mmcls.models <api/models>
+   mmcls.models.utils <api/models.utils>
+   mmcls.datasets <api/datasets>
+   数据转换 <api/transforms>
+   批数据增强 <api/models.utils.augment>
+   mmcls.utils <api/utils>
 
 
 .. toctree::

docs/zh_CN/tools/onnx2tensorrt.md

@@ -2,13 +2,12 @@
 
 <!-- TOC -->
 
-- [ONNX 转 TensorRT（试验性的）](#onnx-tensorrt)
+- [如何将模型从 ONNX 转换到 TensorRT](#如何将模型从-onnx-转换到-tensorrt)
-  - [如何将模型从 ONNX 转换到 TensorRT](#id1)
+  - [准备工作](#准备工作)
-    - [准备工作](#id2)
+  - [使用方法](#使用方法)
-    - [使用方法](#id3)
+- [支持转换至 TensorRT 的模型列表](#支持转换至-tensorrt-的模型列表)
-  - [支持转换至 TensorRT 的模型列表](#tensorrt)
+- [提示](#提示)
-  - [提示](#id4)
+- [常见问题](#常见问题)
-  - [常见问题](#id5)
 
 <!-- TOC -->
 

docs/zh_CN/tools/pytorch2onnx.md

@@ -2,13 +2,12 @@
 
 <!-- TOC -->
 
-- [Pytorch 转 ONNX （试验性的）](#pytorch-onnx)
+- [如何将模型从 PyTorch 转换到 ONNX](#如何将模型从-pytorch-转换到-onnx)
-  - [如何将模型从 PyTorch 转换到 ONNX](#id1)
+  - [准备工作](#准备工作)
-    - [准备工作](#id2)
+  - [使用方法](#使用方法)
-    - [使用方法](#id3)
+- [支持导出至 ONNX 的模型列表](#支持导出至-onnx-的模型列表)
-  - [支持导出至 ONNX 的模型列表](#onnx)
+- [提示](#提示)
-  - [提示](#id4)
+- [常见问题](#常见问题)
-  - [常见问题](#id5)
 
 <!-- TOC -->
 

docs/zh_CN/tools/pytorch2torchscript.md

@@ -2,11 +2,10 @@
 
 <!-- TOC -->
 
-- [Pytorch 转 TorchScript （试验性的）](#pytorch-torchscript)
+- [如何将 PyTorch 模型转换至 TorchScript](#如何将-pytorch-模型转换至-torchscript)
-  - [如何将 PyTorch 模型转换至 TorchScript](#id1)
+  - [使用方法](#使用方法)
-    - [使用方法](#id2)
+- [提示](#提示)
-  - [提示](#id3)
+- [常见问题](#常见问题)
-  - [常见问题](#id4)
 
 <!-- TOC -->
 

docs/zh_CN/tutorials/runtime.md

@@ -7,14 +7,8 @@
 - [定制工作流](#定制工作流)
 - [钩子](#钩子)
   - [默认训练钩子](#默认训练钩子)
-    - [权重文件钩子（CheckpointHook）](#权重文件钩子（checkpointhook）)
-    - [日志钩子（LoggerHooks）](#日志钩子（loggerhooks）)
-    - [验证钩子（EvalHook）](#验证钩子（evalhook）)
   - [使用内置钩子](#使用内置钩子)
   - [自定义钩子](#自定义钩子)
-    - [1. 创建一个新钩子](#1.-创建一个新钩子)
-    - [2. 注册新钩子](#2.-注册新钩子)
-    - [3. 修改配置](#3.-修改配置)
 - [常见问题](#常见问题)
 
 <!-- TOC -->
@@ -183,7 +177,7 @@ custom_hooks = [
 
 ## 自定义钩子
 
-### 1. 创建一个新钩子
+### 创建一个新钩子
 
 这里举一个在 MMClassification 中创建一个新钩子，并在训练中使用它的示例：
 
@@ -218,7 +212,7 @@ class MyHook(Hook):
 
 根据钩子的功能，用户需要指定钩子在训练的每个阶段将要执行的操作，比如 `before_run`，`after_run`，`before_epoch`，`after_epoch`，`before_iter` 和 `after_iter`。
 
-### 2. 注册新钩子
+### 注册新钩子
 
 之后，需要导入 `MyHook`。假设该文件在 `mmcls/core/utils/my_hook.py`，有两种办法导入它：
 
@@ -238,7 +232,7 @@ __all__ = ['MyHook']
 custom_imports = dict(imports=['mmcls.core.utils.my_hook'], allow_failed_imports=False)
 ```
 
-### 3. 修改配置
+### 修改配置
 
 ```python
 custom_hooks = [

requirements/docs.txt

@@ -1,6 +1,6 @@
-docutils==0.16.0
+docutils==0.17.1
 myst-parser
-git+https://github.com/open-mmlab/pytorch_sphinx_theme.git#egg=pytorch_sphinx_theme
+-e git+https://github.com/open-mmlab/pytorch_sphinx_theme.git#egg=pytorch_sphinx_theme
-sphinx==4.0.2
+sphinx==4.5.0
 sphinx-copybutton
 sphinx_markdown_tables