ml_with_django is a opensource template for serving machine learning model with django application. This project also contains a almost production-ready admin dashboard, which is based on django-admin.
You can use this template for developing django based ml-application very quickly only with several steps.
This project uses tensorflow 1.8 version, which still relies on old api version. Update to current tensorflow version may produce incompatible conflict.
Contents of this document
Moved to settings.
- python 3.6.5
- pip
- virtualenv or virtualwrapper
By default setting Django will create a local sqllite.db and use this database for local development.
Create a superuser account, and start application:
$ mkvirtualenv <project_name> # create a new python virutal environment $ workon <project_name> # active this environment $ pip install -r requirements # install project dependencies -> it will take a few minutes, depends on your network... $ python manage.py makemigrations $ python manage.py migrate # create admin tables $ python manage.py createsuperuser --settings=config.settings.local # create local admin user -> enter username, email and password here! $ python manage.py runserver 127.0.0.1:8000 # start local service
For creating new user account, just sign up with admin account and create a new group by using /admin UI and add new user to this group.
Once you submit it, the user should be verified and ready to use.
It contains main program logic. All of packages here must follow MTV-Design-Pattern (Model, Template, View).
users provides a customized user-login and user management
image_inf is a example program for car damage classification ref
This folder contains django url, wsgi settings.<br> You can inherit base.py for test or production deployment scope.
third party program and plugins
Place for storing images, audio, video files on your local filesystem. Subfolder models contains pretrained ML-Models.
Place for storing frontend template
Place for dependency settings.
stores frontend static template like css, html, js for webserver.
$ python manage.py collectstatic
To run the tests, check your test coverage, and generate an HTML coverage report:
$ coverage run manage.py test $ coverage html $ open htmlcov/index.html
$ py.test
Sentry is an error logging aggregator service. You can sign up for a free account at https://sentry.io/signup/?code=cookiecutter or download and host it yourself. The system is setup with reasonable defaults, including 404 logging and integration with the WSGI application.
You must set the DSN url in production.
