TImLab provides an easy to use machine learning training image labeling web platform. It differs from LabelMe by emphasing simplicity and limiting labeling to whole picture instead of its parts.
- Check out the source code.
- Adjust src/timlab/settings.py to suit your needs. See Django documentation for details.
- cd to src folder
- Install Django (if you don't have one):
pip install -r requirements.txt - Create initial migrations:
./manage.py makemigrations projects images - Apply migrations to the DB:
./manage.py migrate - Create superuser of installation:
./manage.py createsuperuser - Collect static files for web serving:
./manage.py collectstatic - Set up Apache or any other web server - read Django documentation for details.
For testing purposes a built-in server will be enough:
./manage.py runserverhttp://localhost:8000
- doc/ - documentation
- src/ - actual code
- src/project/ - training projects and their label lists
- src/image/ - images and their handling functions
- src/timlab/ - main settings