ImageChipper.jl

Overview

Image chipping

faux_image = rand( 256, 256 )
# chip( image/Matrix, ( Chip width, Chip height ), ( Overlap X, Overlap Y ) )
chipped = chip( faux_image, ( 64, 64 ), ( 0.0, 0.0 ) )

where the returned array is a 3-tensor of size ( Chip width, Chip height, Chip # )

Image chipping with labels

images with associated bounding boxes may also be chipped with known ObjectDetectionStats.jl bounding boxes.

faux_image = rand( 256, 256 )

boxes = [   Box(   1,   1,  22,  22 ),
            Box(  55,  55, 222, 222 ),
            Box( 222, 222, 264, 264 )   ]
IoU_thresh = 0.0
chips, boxassignments = chip_with_boxes( faux_image, boxes, ( 64, 64 ) ( 0.0, 0.0 ), IoU_thresh )

Enjoy!