ML workflow

[meghalithic]

1. Match images to their metadata by running "metadata_workflow.R"
2. Create a folder of images
   • name the folder with "-jpg" at end
   • Move all files (.txt and .tif)
   • Remove all files with ending ".txt" by
     • cd to directory
     • rm *.txt
3. Convert to jpg
   • cd to folder
   • for f in *.tif; do  echo "Converting $f"; convert "$f"  "$(basename "$f" .tif).jpg"; done
   • rm *.tif
4. Run DeepBryo for Microporella
   • cd to repository DeepBryo_micro

     
     conda activate deepbryo_micro
     pip3 install openmim
     mim install mmcv-full==1.4.0
     #git clone https://github.com/agporto/DeepBryo.git
     #cd DeepBryo_micro
     pip3 install -e .```

   • cd app && streamlit run app.py --server.port 8080
   • the app can be accessed through the web browser at: localhost:8080
5. Run ML-morph for Microporella
   • cd to repository ml-morph_micro
   • follow instructions here
6. View images and landmarks
   • ./imglab output.xml
     • click + press "i" to ignore
     • MUST SAVE AS NEW FILE TO RETAIN CHANGES!!!!
   • Save as a csv
     • python
     • from utils import dlib_xml_to_pandas
       • ignore apex is not installed
     • dlib_xml_to_pandas('output.xml')
       • must have ".xml" ending
     • quit()