%load_extautoreload%autoreload2%matplotlibinline# # ,o888888o. 8 8888 88 8 8888 8 888888888o. 8 8888888888 8 888888888o. # 8888 `88. 8 8888 88 8 8888 8 8888 `^888. 8 8888 8 8888 `^888. # ,8 8888 `8. 8 8888 88 8 8888 8 8888 `88. 8 8888 8 8888 `88. # 88 8888 8 8888 88 8 8888 8 8888 `88 8 8888 8 8888 `88 # 88 8888 8 8888 88 8 8888 8 8888 88 8 888888888888 8 8888 88 # 88 8888 8 8888 88 8 8888 8 8888 88 8 8888 8 8888 88 # 88 8888 8888888 8 8888 88 8 8888 8 8888 ,88 8 8888 8 8888 ,88 # `8 8888 .8' ` 8888 ,8P 8 8888 8 8888 ,88' 8 8888 8 8888 ,88' # 8888 ,88' 8888 ,d8P 8 8888 8 8888 ,o88P' 8 8888 8 8888 ,o88P' # `8888888P' `Y88888P' 8 8888 8 888888888P' 8 888888888888 8 888888888P' ## P R O O F R E A D I N G# http://github.com/VCG/guidedproofreading#importgp

## LOAD GP MODEL#g=gp.GP() # model='mouse' by default, model='fruitfly'...

Loaded GP model for mouse brain.

## LOAD TEST DATA (AC4 Subvolume 400x400x10vx)## Haehn D, Knowles-Barley S, Roberts M, Beyer J, Kasthuri N, Lichtman JW, Pfister H. # Design and Evaluation of Interactive Proofreading Tools for Connectomics.# IEEE Transactions on Visualization and Computer Graphics 2014;20(12):2466-2475.##image, prob, gold, segmentation=gp.Util.load(0) # <-- slice 0 only, we will load all later..gp.Util.show(image, prob, gold, segmentation)

## RANK SEGMENT PAIR FOR SPLIT ERROR#label1=2365label2=975# Show the pairgp.Util.show_labels(image, segmentation, [label1, label2])
# Rank the pair# 0: correct split# ..# 1: split errorrank=g.rank(image, prob, segmentation, label1, label2)
printrank

0.991059184074

## FIND MERGE ERROR IN SEGMENT #label1=583# Generate potential boundaries in a segment and rank them.# result: the fixed merge error# rank:# 0: uncertain fix# ..# 1: confident fixresult, rank=g.find_merge_error(image, prob, segmentation, label1)
printrank# Show the label and the fixgp.Util.show_labels(image, segmentation, [label1], result)

0.999999998746

#### CORRECT A WHOLE VOLUME###

## LOAD TEST DATA#images, probs, golds, segmentations=gp.Util.load_all() # <-- this time we load all 10 slicesgp.Util.show(images, probs, golds, segmentations)

Loading..
100%

## TODO! (see http://github.com/VCG/gp for all the stuff but it needs some cleanup!)#