Fwd: Re: Run_cytoseg_on_cluster

[Rick Giuly]

-------- Original Message --------
Subject: Re: Run_cytoseg_on_cluster
Date: Tue, 26 Jun 2012 11:40:39 -0700
From: Rick Giuly <rgi...@gmail.com>
To: vadim astakhov <vadim_a...@hotmail.com>
CC: Willy W. Wong <waw...@sdsc.edu>


Hi Vadim,

Here's how to handle step one. (There's also a step2 of the process, but
I'd recommend just testing with step1 first.)


From the command line:

Note the voxelProcessing... parameter

This would do from slice z1 to z2.

> python run_pipeline_test.py data\example\input output
> --trainingImage=data\example\train_images
> --trainingSeg=data\example\train_seg
> --voxelTrainingLowerBound=*,*,*
> --voxelTrainingUpperBound=*,*,*
> --voxelProcessingLowerBound=*,*,z1
> --voxelProcessingUpperBound=*,*,z2
> --contourTrainingLowerBound=*,*,*
> --contourTrainingUpperBound=*,*,*
> --contourProcessingLowerBound=*,*,*
> --contourProcessingUpperBound=*,*,*
> --accuracyCalcLowerBound=*,*,*
> --accuracyCalcUpperBound=*,*,*
> --labelConfigFile=settings3.py
> --voxelWeights=0.0130,0.00064
> --contourListWeights=7,1
> --contourListThreshold=0.8 --step1





From python code:

Looking at
http://code.google.com/p/cytoseg/source/browse/trunk/cytoseg/batch_process.py

for an example

"regionToClassify" specifies the region to process in this code snippet:



regionToClassify = Box([f.cornerA[0], f.cornerA[1], zOffset],
[f.cornerB[0], f.cornerB[1], finalZ])

segmentationParams['steps'] = steps #'classifyVoxels'
segmentationParams['regionToClassify'] = regionToClassify

runSteps(**segmentationParams)





-Rick


vadim astakhov wrote:
>
> Hi Rick.
>
> Could you please give me an example of command line to run sub-set of
> slices. You've mentioned it can be done.
>
>
>
> ...... Cytoseg already knows how to open a subset of slices at a time,
> and processes them in a batch fashion. The code would need a some MPI or
> equivalent worked in to enable parallel execution.
> Installation:http://code.google.com/p/cytoseg/wiki/Installation
> Usage:http://code.google.com/p/cytoseg/wiki/Usage
> -rick
>
>
>
>
> I'd like to try Map and Reduce approach. But example from "Usage" does
> not show how to do that. Thanks a lot in advance.
>
>
> python run_pipeline_test.py data\example\input output
> --trainingImage=data\example\train_images
> --trainingSeg=data\example\train_seg
> --voxelTrainingLowerBound=*,*,*
> --voxelTrainingUpperBound=*,*,*
> --voxelProcessingLowerBound=*,*,*
> --voxelProcessingUpperBound=*,*,*
> --contourTrainingLowerBound=*,*,*
> --contourTrainingUpperBound=*,*,*
> --contourProcessingLowerBound=*,*,*
> --contourProcessingUpperBound=*,*,*
> --accuracyCalcLowerBound=*,*,*
> --accuracyCalcUpperBound=*,*,*
> --labelConfigFile=settings3.py
> --voxelWeights=0.0130,0.00064
> --contourListWeights=7,1
> --contourListThreshold=0.8 --step1--step2
>
>
>
>
>
>
>
>
>