Issues reusing topaz trained models

Data Management
1

I had exported a topaz model I trained (Job 13 - J13) as a group. After I import the model in another session, cryosparc looks for the files in folder J13 when topaz is launched. cryosparc also saves the preprocessed files in J13.

Is there a way for this to not require the J13 folder in a new/different sessions? Is there another way to reuse my own trained models?
To be able to use my own pre-trained model is a bit of a production.

  1. deleting J13 in new project.
  2. recreating J13 with the trained model.
  3. importing the model
  4. running topaz extract until it preprocesses everything into J13 and crashes
  5. relaunching topaz with the new preprocessed folder now located in J13
2

Hi @August,

Thanks for question! You should be able to import the model itself using the “Import Result Group” job and specifying the location of the .csg file associated with your TOPAZ model (eg. /path/to/CS-project/J13/J13_topaz_model.csg). Then in your new project, make a new directory (I usually call it “topaz_preprocess”) and point the TOPAZ extract job to look in that folder for the preprocessed mics (it will be empty, but TOPAZ will then preprocess all the new micrographs).

Please let me know if this helps.