A tool for images markup for deep learning puproses.
MVP: Color segmentation masks.
Interface model:
- Line "Input directory" display absolute path to folder with input images. To change it press "Change" button or edit path manually.
- Line "Output directory" display absolute path to folder where segmentation mask was saved. To change it press "Change" button or edit path manually.
- User able to switch between 2 instruments: brush and polygon.
- Size of brush can be changed by Ctrl+mouse wheel. Size of brush cursor always fits the width of drawing line.
- Polygon item places vertecies on image by mouse clicking. If user press on first-placed vertex, tool draw a polygon with marked verticies. Other verticies can be moved in the markup process by holding mouse button.
- User able to choose markup color on color Palette. Application provides 12 different colors.
- User press on image name in list on the left and it is loaded on the screen. If canvas have unsaved changes, application suggest save them before switching image.
- Button "Save" creates the file with a mask in the output directory. File has the same name as original image has.
- Button "Clear" remove mark objects from image.
- Buttons "Undo", "Redo" allow user to move back and forth along markup history.
-
Download, install and setup Git LFS
-
Create and setup virtualenv
python3 -m venv /path/to/new/virtual/environment
source <venv>/bin/activate
pip install -r requirements.dev.txt
- Install package
pip install -e .
Now you can run app with
dl_markup
pytest --cov=dl_markup
flake8 dl_markup
pydocstyle dl_markup
pylupdate5 dl_markup/*.py -ts dl_markup.ru.ts
lrelease dl_markup.ru.ts
python -m build .
This command generate two files in the dist
directory:
dist\
dl_markup-0.0.1-py3-none-any.whl
dl_markup-0.0.1.tar.gz