http://imagej.273.s1.nabble.com/Trainable-Segmentation-tp5002107p5002150.html
My understanding is that you want to use Trainable segmentation not necessarily to develop it. Trainable segmentation is one of the Fiji specific projects that rely on Fiji's facilities. The people who developed it (Ignacio) have their day jobs and probably don't have the time and energy to make it run under ImageJ.
Why don't you just try to download Fiji and run it on your data? It will certainly require less effort then making it run seamlessly in ImageJ.
If trainable segmentation is what you and your base of biology users want to use and insist on running it under ImageJ then we can discuss whether such migration is possible. A viable alternative may be to have both ImageJ and Fiji installed and revert to Fiji when trainable segmentation is required.
-----------------------------------------------------------------------------------
Pfotenhauerstr. 108
-----------------------------------------------------------------------------------
> Both you and Mr. Schindelin gave very similar answers to my question--answers which, as a regular reader of the group, I expected. I was also hoping to hear something along the lines of "Go to x and download the plugin, it's clean of any other dependencies", but that does not seem to be true in this case. I'll try to address your question, though. There are good reasons, from my perspective, to choose ImageJ over FIJI.
>
> 1) I've already got an installed base of half a dozen non-programmer users using ImageJ, with a bunch of plug-ins installed, many of which I wrote, and many of which I didn't. Switching would involve the hassle of converting all their machines (personally--they aren't savvy enough to do it themselves), and then reinstalling all the plug-ins under FIJI as well, although some might already be there.
>
> 2) I'm not a real programmer--I just play one at work. All the fancy tools real programmers use are a barrier to entry for me. Don't get me wrong--Eclipse, Github, Netbeans etc. are all great tools for those who code for a living. But I spend maybe 15% of my time programming at work. I'll go months without writing a line of code, until a problem comes along that needs it. Then it's critically important! The rest of the time I'm in the lab, at meetings, juggling budgets, etc. Even if all those tools cut down my programming time by half, it would take years to recover the time invested in learning them because I don't code enough to make it worthwhile. I have other areas of expertise that are more critical to my job that I invest my training time in.
>
> 3) So I have to stick to simple to learn. Formal training is out of the question. Every language I currently use (Java, macro, Visual Basic, a smattering of JavaScript) is self-taught. Notepad++ to edit. "Compile and run" in ImageJ, and hope any error messages aren't too mysterious. Hand coded build files to make .jars for the more complex stuff. My "code repository" is a folder full of .java files (backed up offsite, naturally). Need a plugin? Pull it off the ImageJ site and dump it in the plugins folder. The code is at the same place, or often right in the .jar! Marvelously simple. But my projects are small enough that I can get away with that. ImageJ is a perfect match for this style. Yes, FIJI can be used the same way. But the spirit of it seems to be more oriented towards using real programming tools, and making sure everything fits correctly, etc. Great for efficiency but again that barrier thing. So the short answer to the question "Why ImageJ instead of FIJI?" is, in my case, ImageJ has everything I need and less that I don't. FIJI seems to add a lot of stuff I don't need without adding anything I do (well, except maybe Trainable Segmentation...).
>
> I guess the purpose of this lengthy response is a gentle reminder to the experts that not everybody else can be. Where possible, suggest solutions that require little or no specialized knowledge. Sometimes quick and dirty really is better than elegant and bullet proof, if it's a lot simpler. I can't be an expert in everything. My plumber, accountant, and karate instructor all tell me I could be so much better at what they do if I just devoted the effort to it that they do...but none of them have a degree in my field, either. So most of the people asking questions on the list are not only less of a programming expert than the FIJI team, they also are less likely to be able to invest the time and energy to become one.
>
> For the case at hand, I'll download FIJI just to try out the Trainable Segmentation. If it handles my problem, great! Then I'll need to decide if it's easier to switch everybody to FIJI or come up the learning curve (it might not be very big...I haven't looked) on using a real code repository so I can pull what I need out of it and drop it into ImageJ. Is the stuff in the repository in .jar files and ready to install, or would I also need to rebuild it? I'll figure it out if I need to.
>
> Thanks for letting me ramble. And thank you (both the ImageJ and FIJI teams), not only for your help, but for all the time and energy you spend on making an incredibly useful tool for the rest of us to use--even if we can't master the intricacies it takes to build it.
>
> Bob
> -----Original Message----- From: Ignacio Arganda-Carreras
> Sent: Monday, March 11, 2013 10:17 PM
> To:
[hidden email]
> Subject: Re: Trainable Segmentation
>
> Dear Bob,
>
> You can download the Trainable Segmentation package (including
> Trainable_Segmentation and Weka_Segmentation plugins) and all its
> dependencies from the Fiji repository. But why don't you use it within Fiji
> so you don't have to worry about the dependencies?
>
> ignacio
>
> On Mon, Mar 11, 2013 at 8:46 PM, Bob <
[hidden email]> wrote:
>
>> Recently the Trainable Segmentation plugin in FIJI was mentioned in
>> another thread. It looks very interesting for one of my applications. Is
>> it available as a plugin for ImageJ? If so, where can I download the .jar?
>> Are there other plug-ins that it depends on that I’d also need?
>>
>> Thank you!
>>
>> --
>> ImageJ mailing list:
http://imagej.nih.gov/ij/list.html>>
>
>
>
> --
> Ignacio Arganda-Carreras, Ph.D.
> Seung's lab, 46-5065
> Department of Brain and Cognitive Sciences
> Massachusetts Institute of Technology
> 43 Vassar St.
> Cambridge, MA 02139
> USA
>
> Phone: (001) 617-324-3747
> Website:
http://bioweb.cnb.csic.es/~iarganda/index_EN.html>
> --
> ImageJ mailing list:
http://imagej.nih.gov/ij/list.html
> --
> ImageJ mailing list:
http://imagej.nih.gov/ij/list.html