generateSuperpixels

Implementation of various superpixel algorithms for generating over-segmentations of an image. Algorithms include:

• SLIC superpixels from Achanta et al., PAMI 2012.
• graph-cut superpixels from Zhang et al., ICCV 2011.

Alternatively, can produce non-contiguous superpixels using k-means clustering (see the -m option).

Example:

bin/generateSuperpixels -g 100 -o <OUTPUT_SP_NAME> -x <IMAGE_FILENAME>
bin/visualizeSuperpixels -x -i <IMAGE_FILENAME> <INPUT_SP_NAME>

Multitple over-segmentations can be written to the same file using the -a option or multiple repeates of -g or -k. For example,

bin/generateSuperpixels -m SUPERPIXEL -g 10 -g 100 -o <OUTPUT_SP_NAME> <IMAGE_FILENAME>
bin/generateSuperpixels -m SLIC -g 10 -g 100 -a <OUTPUT_SP_NAME> <IMAGE_FILENAME>
bin/visualizeSuperpixels -x -i <IMAGE_FILENAME> <INPUT_SP_NAME>

See Also

drwnSLICSuperpixels

drwnFastSuperpixels

drwnKMeansSegments