org.datavec.image.loader

Class ImageLoader

java.lang.Object

org.datavec.image.loader.BaseImageLoader

org.datavec.image.loader.ImageLoader

All Implemented Interfaces:

Serializable

public class ImageLoader
extends BaseImageLoader

Image loader for taking images and converting them to matrices

Author:

Adam Gibson

See Also:

Serialized Form

Field Summary

Fields inherited from class org.datavec.image.loader.BaseImageLoader

ALLOWED_FORMATS, BASE_DIR, centerCropIfNeeded, channels, height, imageTransform, log, rng, width

Constructor Summary

Constructors

Constructor and Description
ImageLoader()
ImageLoader(int height, int width) Instantiate an image with the given height and width
ImageLoader(int height, int width, int channels) Instantiate an image with the given height and width
ImageLoader(int height, int width, int channels, boolean centerCropIfNeeded) Instantiate an image with the given height and width

Method Summary

Modifier and Type	Method and Description
org.nd4j.linalg.api.ndarray.INDArray	asImageMiniBatches(File f, int numMiniBatches, int numRowsPerSlice) Slices up an image in to a mini batch.
org.nd4j.linalg.api.ndarray.INDArray	asMatrix(BufferedImage image) Convert an BufferedImage to a matrix
org.nd4j.linalg.api.ndarray.INDArray	asMatrix(File f) Convert an image file in to a matrix
org.nd4j.linalg.api.ndarray.INDArray	asMatrix(InputStream inputStream) Convert an input stream to a matrix
org.nd4j.linalg.api.ndarray.INDArray	asRowVector(BufferedImage image) Convert an image in to a row vector
org.nd4j.linalg.api.ndarray.INDArray	asRowVector(File f) Convert a file to a row vector
org.nd4j.linalg.api.ndarray.INDArray	asRowVector(InputStream inputStream)
BufferedImage	centerCropIfNeeded(BufferedImage img)
int[]	flattenedImageFromFile(File f)
int[][]	fromFile(File file) Load a rastered image from file
int[][][]	fromFileMultipleChannels(File file) Load a rastered image from file
protected BufferedImage	scalingIfNeed(BufferedImage image, boolean needAlpha)
protected BufferedImage	scalingIfNeed(BufferedImage image, int dstHeight, int dstWidth, boolean needAlpha)
org.nd4j.linalg.api.ndarray.INDArray	toBgr(BufferedImage image) Convert an BufferedImage to an bgr spectrum image
org.nd4j.linalg.api.ndarray.INDArray	toBgr(File file) Convert an input stream to an bgr spectrum image
org.nd4j.linalg.api.ndarray.INDArray	toBgr(InputStream inputStream) Convert an input stream to an bgr spectrum image
static BufferedImage	toBufferedImage(Image img, int type) Converts a given Image into a BufferedImage
void	toBufferedImageRGB(org.nd4j.linalg.api.ndarray.INDArray arr, BufferedImage image) Convert the given image to an rgb image
static BufferedImage	toImage(org.nd4j.linalg.api.ndarray.INDArray matrix) Convert a matrix in to a buffereed image
protected org.nd4j.linalg.api.ndarray.INDArray	toINDArrayBGR(BufferedImage image)
protected int[][]	toIntArrayArray(BufferedImage image)
org.nd4j.linalg.api.ndarray.INDArray	toRaveledTensor(BufferedImage image) Convert an image in to a raveled tensor of the bgr values of the image
org.nd4j.linalg.api.ndarray.INDArray	toRaveledTensor(File file) Changes the input stream in to an bgr based raveled(flattened) vector
org.nd4j.linalg.api.ndarray.INDArray	toRaveledTensor(InputStream is) Changes the input stream in to an bgr based raveled(flattened) vector

Methods inherited from class org.datavec.image.loader.BaseImageLoader

downloadAndUntar, getAllowedFormats

Methods inherited from class java.lang.Object

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

Constructor Detail

ImageLoader

public ImageLoader()

ImageLoader

public ImageLoader(int height,
                   int width)

Instantiate an image with the given height and width

Parameters:

height - the height to load*

width - the width to load

ImageLoader

public ImageLoader(int height,
                   int width,
                   int channels)

Instantiate an image with the given height and width

Parameters:

height - the height to load

width - the width to load

channels - the number of channels for the image*

ImageLoader

public ImageLoader(int height,
                   int width,
                   int channels,
                   boolean centerCropIfNeeded)

Instantiate an image with the given height and width

Parameters:

height - the height to load

width - the width to load

channels - the number of channels for the image*

centerCropIfNeeded - to crop before rescaling and converting

Method Detail

asRowVector

public org.nd4j.linalg.api.ndarray.INDArray asRowVector(File f)
                                                 throws IOException

Convert a file to a row vector

Specified by:

asRowVector in class BaseImageLoader

Parameters:

f - the image to convert

Returns:

the flattened image

Throws:

IOException

asRowVector

public org.nd4j.linalg.api.ndarray.INDArray asRowVector(InputStream inputStream)
                                                 throws IOException

Specified by:

asRowVector in class BaseImageLoader

Throws:

IOException

asRowVector

public org.nd4j.linalg.api.ndarray.INDArray asRowVector(BufferedImage image)

Convert an image in to a row vector

Parameters:

image - the image to convert

Returns:

the row vector based on a rastered representation of the image

toRaveledTensor

public org.nd4j.linalg.api.ndarray.INDArray toRaveledTensor(File file)

Changes the input stream in to an bgr based raveled(flattened) vector

Parameters:

file - the input stream to convert

Returns:

the raveled bgr values for this input stream

toRaveledTensor

public org.nd4j.linalg.api.ndarray.INDArray toRaveledTensor(InputStream is)

Changes the input stream in to an bgr based raveled(flattened) vector

Parameters:

is - the input stream to convert

Returns:

the raveled bgr values for this input stream

toRaveledTensor

public org.nd4j.linalg.api.ndarray.INDArray toRaveledTensor(BufferedImage image)

Convert an image in to a raveled tensor of the bgr values of the image

Parameters:

image - the image to parse

Returns:

the raveled tensor of bgr values

toBgr

public org.nd4j.linalg.api.ndarray.INDArray toBgr(File file)

Convert an input stream to an bgr spectrum image

Parameters:

file - the file to convert

Returns:

the input stream to convert

toBgr

public org.nd4j.linalg.api.ndarray.INDArray toBgr(InputStream inputStream)

Convert an input stream to an bgr spectrum image

Parameters:

inputStream - the input stream to convert

Returns:

the input stream to convert

toBgr

public org.nd4j.linalg.api.ndarray.INDArray toBgr(BufferedImage image)

Convert an BufferedImage to an bgr spectrum image

Parameters:

image - the BufferedImage to convert

Returns:

the input stream to convert

asMatrix

public org.nd4j.linalg.api.ndarray.INDArray asMatrix(File f)
                                              throws IOException

Convert an image file in to a matrix

Specified by:

asMatrix in class BaseImageLoader

Parameters:

f - the file to convert

Returns:

a 2d matrix of a rastered version of the image

Throws:

IOException

asMatrix

public org.nd4j.linalg.api.ndarray.INDArray asMatrix(InputStream inputStream)
                                              throws IOException

Convert an input stream to a matrix

Specified by:

asMatrix in class BaseImageLoader

Parameters:

inputStream - the input stream to convert

Returns:

the input stream to convert

Throws:

IOException

asMatrix

public org.nd4j.linalg.api.ndarray.INDArray asMatrix(BufferedImage image)

Convert an BufferedImage to a matrix

Parameters:

image - the BufferedImage to convert

Returns:

the input stream to convert

asImageMiniBatches

public org.nd4j.linalg.api.ndarray.INDArray asImageMiniBatches(File f,
                                                               int numMiniBatches,
                                                               int numRowsPerSlice)

Slices up an image in to a mini batch.

Parameters:

f - the file to load from

numMiniBatches - the number of images in a mini batch

numRowsPerSlice - the number of rows for each image

Returns:

a tensor representing one image as a mini batch

flattenedImageFromFile

public int[] flattenedImageFromFile(File f)
                             throws IOException

Throws:

IOException

fromFile

public int[][] fromFile(File file)
                 throws IOException

Load a rastered image from file

Parameters:

file - the file to load

Returns:

the rastered image

Throws:

IOException

fromFileMultipleChannels

public int[][][] fromFileMultipleChannels(File file)
                                   throws IOException

Load a rastered image from file

Parameters:

file - the file to load

Returns:

the rastered image

Throws:

IOException

toImage

public static BufferedImage toImage(org.nd4j.linalg.api.ndarray.INDArray matrix)

Convert a matrix in to a buffereed image

Parameters:

matrix - the

Returns:

BufferedImage

toBufferedImageRGB

public void toBufferedImageRGB(org.nd4j.linalg.api.ndarray.INDArray arr,
                               BufferedImage image)

Convert the given image to an rgb image

Parameters:

arr - the array to use

image - the image to set

toBufferedImage

public static BufferedImage toBufferedImage(Image img,
                                            int type)

Converts a given Image into a BufferedImage

Parameters:

img - The Image to be converted

type - The color model of BufferedImage

Returns:

The converted BufferedImage

toIntArrayArray

protected int[][] toIntArrayArray(BufferedImage image)

toINDArrayBGR

protected org.nd4j.linalg.api.ndarray.INDArray toINDArrayBGR(BufferedImage image)

centerCropIfNeeded

public BufferedImage centerCropIfNeeded(BufferedImage img)

scalingIfNeed

protected BufferedImage scalingIfNeed(BufferedImage image,
                                      boolean needAlpha)

scalingIfNeed

protected BufferedImage scalingIfNeed(BufferedImage image,
                                      int dstHeight,
                                      int dstWidth,
                                      boolean needAlpha)