/tomoSgmt

Vesicle Segmentation for Missing-Wedge Corrected tomograms

Primary LanguagePython

Example Code

####Predict vesicle from a missing-wedge corrected tomogram:

  1. cd /storage/heng/tomoSgmt/example/test_2021Jan/t208/
  2. mkdir ppXXXX cd ppXXXX
  3. cp /storage/heng/tomoSgmt/bin/getinfo.py ./
  4. python getinfo.py
  5. cd demo
  6. python proc_cmd.py (now model_d4_era40-100_70_1.h5 is the best one, make sure model file in /t208/, now in /bin/)
  7. copy the command and run

then, in demo dir, some useful file will be generated such as (something new listed): missed.mod (prediction result compare with vesicle_area.xml) wrong.mod

#Some scripts new:

  1. random_erase.py used for train data(origin tomo), generate N1 2d squares and N2 3d cubes

  2. relocal_truth.py vesicle.xml file contains vesicles out of area_file, use this script to generate vesicle_area.xml file (contained in getinfo module)

  3. json2mod.py .json file in prediction result(for example, pp0365_vesicle.json) to point.mod

  4. show_all_ves.py vesicle_area.xml to vesall.mod