Output data, returned as a table with MiniBatchSize
number of rows.
The first variable is InputImage
and contains input image
patches.
If the network responses are images in an image datastore, then the second
variable is ResponseImage
.
If the network responses are pixel label images in a pixel label datastore,
then the second variable is ResponsePixelLabelImage
.
Each column contains a cell array of patches of size determined by
PatchSize
and the type of image data.
For 2-D single-channel images, the patches are size
m-byn, where m
specifies the number of rows and n specifies the number of
columns in the patch.
For 2-D multi-channel images, the patches are size
m-byn-by-c, where
c specifies the number of color channels of the patch.
c is 3 for RGB images.
For 3-D single-channel volumetric images, the patches are size
m-byn-by-p, where
p specifies the number of planes of the volume.
For the last batch of data in the datastore patchds
, if
numObservations
is not cleanly divisible by
MiniBatchSize
, then read
returns a partial batch
containing all the remaining observations in the datastore.