Allen Cell Discussion Forum

Iterative Deep Learning Workflow

Hello Allen Segmentation Developers,

I am trying to use the Iterative Deep Learning Workflow on a Windows 10 operating system utilizing a Bash Shell with an Ubuntu command window (Windows Subsystem for Linux) (version 1).

image

I have followed the aics-ml-segmentation / READ.md instructions exactly. At the install Pytorch setting I chose Stable (1.9.0), Linux, Conda, Python, CPU. I cloned everything and I tried to use the curator_sorting help feature. The first error code I received was module not found error: no module named ‘aicssegmentation’.

image

I then pip installed aicssegmentation. I tried to use the curator_sorting help function again and received this error.

image

I can get the Classic Image Segmentation Workflow to work properly on Windows Subsystem for Linux, In addition, we can install the Iterative Deep Learning Workflow on a Mac and can run the curator_sorting help function without an error. However, we can cannot we get the above error on the Ubuntu command window (Windows Subsystem for Linux).

Additional information you might need:

pip show aicssegmentation

image

pip show aicsmlsegment

image

Please advise me on what I might be missing or need to install to make the mlsegmenter environment work.

Best regards,

RexH

Hello @RexH , Thanks for reporting the missing package aicssegmentation in the requirement. I have fixed this in the latest version.

Regarding your errors related to TkAgg, I would suggest you to try running this in Anaconda prompt for Windows. Follow the guide here: Installing on Windows — Anaconda documentation

I just tried one more time curator_sorting -h . It works well within Anaconda prompt.

Let me if you have more issues.

Thanks,
Jianxu

Hello Jianxu_Chen, I want to thank you for your previous post on how to resolve our TKAgg error. Your suggestion worked very well.

I am posting again as I am unable to get the interface to pop up showing specified slices and a max projection within the curator_sorting. I have included screen shots of the images from Fiji that I am using and a screen shot of the Anaconda prompt for Windows. When I press enter this is what happens in the highlighted red rectangle.

image

image

Any suggestions you may have would be welcome.

Best regards,

RexH

Hello @jianxu_chen, I want to thank you for your previous post on how to resolve our TKAgg error. Your suggestion worked very well.

I am reposting a previous question as I am wondering if it might not have alerted you to the post.

I am unable to get the interface to pop up showing specified slices and a max projection within the curator_sorting. I have included screen shots of the images from Fiji that I am using and a screen shot of the Anaconda prompt for Windows. When I press enter this is what happens in the highlighted red rectangle.

image
image

Any suggestions you may have would be welcome.

Best regards,

RexH

Dear @RexH , Sorry for the late reply. This looks very strange. I will try to reproduce this error on my end and get back to you asap.

Thanks,
Jianxu

hi @RexH , May I ask what will you see in the curator_sorting_record.csv? Also, can you double check the raw .tiff files exist under the directory you used for --raw_path?

Hi @jianxu_chen,

This is the output to the curator_sorting_record.csv

This is inside the directory for the --raw_path.

image

I am sending you an image of the properties just to verify that it is a .tiff image.

Thank you,

RexH

Looks like your file extention is .tif instead of .tiff

So, all you need to do is to change --data_type to .tif

Hi @jianxu_chen,

I have changed all the files to all be .tif as the batch_processing from the classic segmenter creates .tiff files. Now when I run the curator_sorting I receive these error codes.

This is different than what I was receiving before which was

image

Regards,

RexH

Hi @RexH , So sorry if my previous message was confusing. To be clear

  • DO not change the segmentation files. .tiff is correct
  • When you type in the command, one of the things you type in is --data_type, currently, your command has --data_type .tiff. Since you raw images are .tif, you need to do --data_type .tif

Thanks,
Jianxu

Hi @jianxu_chen,

Thank you so much for your help! It is working now. I was very happy to see the interface populate. Moving on to the next step of the process.

Regards,

RexH