request suggestion for weka trainable Segmentation

[zhaoliang]

Hi all,

I meet some problem when trying Weka trainable Segmentation.

My image is 913MB, having 3.63E7 * 3.29E6 Pixels. Weka plugin cannot work well and it shows


Error when updating feature stack.
Feature stack was not updated.
The training did not finish.

I am not sure whether it is because of my huge image that FIji cannot process or any other problem.
My computer is linux-64 bit. The memory/Threads is more than 14GB.

By the way, what I want is make segmentation of the different porosity in the image, is there any better methods to do that in this monstor? Cheers!

Any suggestion please? Cheers!

Best,
Zhaoliang

[Ignacio Arganda-Carreras-2]

Dear Zhaoliang,

Your image is indeed too large for your RAM to process all the features
that the Trainable Weka Segmentation (TWS) creates by default. I would
recommend you to try any of the following strategies:

1) Downsample your image to a size you can work with on your machine. Once
you get a a proper result, you can upsample it to the original size. This
approach of course involves loosing resolution but it might be enough
depending on your segmentation task.

2) Cut your image into pieces, train TWS on some representative pieces,
apply the trained classifier to all pieces and recompose the original image.

3) Get a computer with more RAM.

I hope this helps!

ignacio

On Tue, Mar 28, 2017 at 12:49 AM, zhaoliang <[hidden email]> wrote:

> Hi all,
>
> I meet some problem when trying Weka trainable Segmentation.
>
> My image is 913MB, having 3.63E7 * 3.29E6 Pixels. Weka plugin cannot work
> well and it shows
>
> *
> Error when updating feature stack.
> Feature stack was not updated.
> The training did not finish.*
>
> I am not sure whether it is because of my huge image that FIji cannot
> process or any other problem.
> My computer is linux-64 bit. The memory/Threads is more than 14GB.
>
> By the way, what I want is make segmentation of the different porosity in
> the image, is there any better methods to do that in this monstor? Cheers!
>
> Any suggestion please? Cheers!
>
> Best,
> Zhaoliang
>
>
>
>
> --
> View this message in context: http://imagej.1557.x6.nabble.
> com/request-suggestion-for-weka-trainable-Segmentation-tp5018399.html
> Sent from the ImageJ mailing list archive at Nabble.com.
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
Ignacio Arganda-Carreras, Ph.D.
Ikerbasque Research Fellow
Departamento de Ciencia de la Computacion e Inteligencia Artificial
Facultad de Informatica, Universidad del Pais Vasco
Paseo de Manuel Lardizabal, 1
20018 Donostia-San Sebastian
Guipuzcoa, Spain

Phone : [phone]
Website: http://sites.google.com/site/iargandacarreras/
<http://biocomp.cnb.csic.es/~iarganda/index_EN.html>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html