marisn / timlab

Training IMage LABeler

Training IMage LABeler

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.

Installation

1. Check out the source code.
2. Adjust src/timlab/settings.py to suit your needs. See Django documentation for details.
3. cd to src folder
4. Install Django (if you don't have one): pip install -r requirements.txt
5. Create initial migrations: ./manage.py makemigrations projects images
6. Apply migrations to the DB: ./manage.py migrate
7. Create superuser of installation: ./manage.py createsuperuser
8. Collect static files for web serving: ./manage.py collectstatic
9. 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 runserver http://localhost:8000

Directory layout

• doc/ - documentation
• src/ - actual code
• src/project/ - training projects and their label lists
• src/image/ - images and their handling functions
• src/timlab/ - main settings