Package ai.djl.basicdataset.cv.classification

Class Mnist

java.lang.Object
ai.djl.training.dataset.RandomAccessDataset
ai.djl.training.dataset.ArrayDataset
ai.djl.basicdataset.cv.classification.Mnist
All Implemented Interfaces:
ai.djl.training.dataset.Dataset
public final class Mnist extends ai.djl.training.dataset.ArrayDataset
MNIST handwritten digits dataset from http://yann.lecun.com/exdb/mnist.

Each sample is a grayscale image (in 3-D NDArray) with shape (28, 28, 1).

It is a common starting dataset because it is small and can train within minutes. However, it is an overly easy task that even poor models can still perform very well on. Instead, consider FashionMnist which offers a comparable speed but a more reasonable difficulty task.

  • Nested Class Summary

    Nested Classes
    Modifier and Type
    Class
    Description
    static final class 
    Mnist.Builder
    A builder for a Mnist.

    Nested classes/interfaces inherited from class ai.djl.training.dataset.RandomAccessDataset

    ai.djl.training.dataset.RandomAccessDataset.BaseBuilder<T extends ai.djl.training.dataset.RandomAccessDataset.BaseBuilder<T>>

    Nested classes/interfaces inherited from interface ai.djl.training.dataset.Dataset

    ai.djl.training.dataset.Dataset.Usage

  • Field Summary

    Fields
    Modifier and Type
    Field
    Description
    static final int
    IMAGE_HEIGHT
     
    static final int
    IMAGE_WIDTH
     
    static final int
    NUM_CLASSES
     

    Fields inherited from class ai.djl.training.dataset.ArrayDataset

    data, labels

    Fields inherited from class ai.djl.training.dataset.RandomAccessDataset

    dataBatchifier, device, labelBatchifier, limit, pipeline, prefetchNumber, sampler, targetPipeline

  • Method Summary

    Modifier and Type
    Method
    Description
    static Mnist.Builder
    builder()
    Creates a builder to build a Mnist.
    void
    prepare(ai.djl.util.Progress progress)

    Methods inherited from class ai.djl.training.dataset.ArrayDataset

    availableSize, get, getByIndices, getByRange, getData, newSubDataset, newSubDataset

    Methods inherited from class ai.djl.training.dataset.RandomAccessDataset

    getData, getData, getData, randomSplit, size, subDataset, subDataset, subDataset, subDataset, toArray

    Methods inherited from class java.lang.Object

    clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

    Methods inherited from interface ai.djl.training.dataset.Dataset

    matchingTranslatorOptions, prepare

  • Field Details

  • Method Details

    • builder

      public static Mnist.Builder builder()
      Creates a builder to build a Mnist.
      Returns:
      a new builder

    • prepare

      public void prepare(ai.djl.util.Progress progress) throws IOException
      Specified by:
      prepare in interface ai.djl.training.dataset.Dataset
      Overrides:
      prepare in class ai.djl.training.dataset.ArrayDataset
      Throws:
      IOException