mindspore::dataset
This module provides APIs to load and process various common datasets such as MNIST, CIFAR-10, CIFAR-100, VOC, COCO, ImageNet, CelebA, CLUE, etc. It also supports datasets in standard format, including MindRecord, TFRecord, Manifest, etc. Users can also define their own datasets with this module.
Besides, this module provides APIs to sample data while loading.
Notice that cache is not supported on Windows platform yet. Please do not use it while loading and processing data on Windows.
Dataset Functions
Dataset Functions provides various functions to load and process datasets.
Vision
API |
Descriptions |
Extra overload APIs (Parameter Sampler is overloaded) |
|---|---|---|
A source dataset for reading and parsing Album dataset and return an object of Class AlbumDataset. |
||
A source dataset for reading and parsing CelebA dataset and return an object of Class CelebADataset. |
||
A source dataset for reading and parsing Cifar100 dataset and return an object of Class Cifar100Dataset. |
||
A source dataset for reading and parsing Cifar10 dataset and return an object of Class Cifar10Dataset. |
||
A source dataset for reading and parsing Coco dataset and return an object of Class CocoDataset. |
||
A source dataset for reading images from a tree of directories and return an object of Class ImageFolderDataset. |
Function mindspore::dataset::ImageFolder (raw ptr Sampler) Function mindspore::dataset::ImageFolder (reference Sampler) |
|
A source dataset for reading and parsing the MNIST dataset and return an object of Class MnistDataset. |
||
A source dataset for reading and parsing the VOC dataset and return an object of Class VOCDataset. |
Text
API |
Descriptions |
Extra overload APIs |
|---|---|---|
A source dataset for reading and parsing CLUE dataset and return an object of Class CLUEDataset. |
None |
Standard Format
API |
Descriptions |
Extra overload APIs (Parameter Sampler is overloaded) |
|---|---|---|
A source dataset for reading and parsing comma-separated values (CSV) datasets and return an object of Class CSVDataset. |
None |
|
A source dataset for reading images from a Manifest file and return an object of Class ManifestDataset. |
||
A source dataset for reading and parsing single MindRecord file and return an object of Class MindDataDataset. |
||
A source dataset for reading and parsing multiple MindRecord files and return an object of Class MindDataDataset. |
||
A source dataset for generating random data and return an object of Class RandomDataDataset. |
None |
|
A source dataset for reading and parsing datasets stored on disk in text format and return an object of Class TextFileDataset. |
None |
|
A source dataset for reading and parsing datasets stored on disk in TFData format and return an object of Class TFRecordDataset. |
None |
Dataset Classes
Dataset Classes provides the definition of base class of dataset and common transform operations of dataset such as map, shuffle and batch. It also provides the definition of Iterator for fetching data.
Sampler Classes
Sampler Classes provides the definitions of samplers, which are used to choose samples from the dataset.
Eager Classes
Eager Classes provides the definitions of Execute class, which is used to apply transforms (e.g. vision/text) on input tensor in eager mode.
Constants
Constants provides some common enums and const variables.
Others
This section contains some predefined classes related to Dataset operations, tool functions, and some Typedefs.
Classes
Functions
Typedefs
Lite-CV
Lite-CV is a special library contrains image transform methods which are implemented without OpenCV.
Note that this library will only be compiled in lite mode of MindSpore with option -n lite_cv.
With this library, the size of lite package will be smaller compared to other libraries links to OpenCV.