personal image-classifier webapp

Getting Started

It would be a good idea to fork the repository because there is a lot of personal information on the website

things that would need to be changed for your own website:

• change personal information in home.html and cv.html

• include Photos for the frontend

• add email configuration under webapp/static/conf/email_conf.py

  • format: see email_conf_example.py, MAIL_SERVER and MAIL_PORT need to be changed depending on your provider

  • if you get an google api error take a look at this answer

• include your own cv named cv.pdf in ./webapp/documents

Installing

Clone Repo or fork

git clone https://github.com/DollofCuty/deep-learning-webapp.git

How to run

With docker-compose

This includes nginx and letsencrypt setup so you have to do the following steps:

• change environment parameters on gunicorn docker in docker-compose.yml:

  • LETSENCRYPT_EMAIL=[your-email]
  • LETSENCRYPT_HOST=[your-domain-name]
  • VIRTUAL_HOST=[your-domain-name]
  • The domain has to be valid and connected with the instance you are running docker-compose on
  • then for namecheap.com I had to choose custom dns and enter the ns-cloud nameservers (this should be possible for all domain registrars)

• set up a google cloud instance with container-optimized-os

  • also follow the section 'Making an alias to Docker Compose' to shorten the commands for the future

docker network create webproxy

docker-compose up

The first time you run this will take some time because it has to build all the containers (web, nginx-web, nginx-gen, nginx-letsencrypt) and then nginx-letsencrypt will request a certificate

If all went well the webservice should use nginx and the letsencrypt certificate to run on https, with reverse proxy and serving static files

updating the docker-compose setup (when you change the Dockerfile for example)

docker-compose build

freeing disk space if you have dangling/unnecassary images (this removes all unused images!)

for reference look at the docker docs

docker system prune -a

With docker (Linux, Mac, Windows)

This does not include nginx and https connection!

docker build -t webapp:latest .

docker run -p 8080:8080 webapp

Flask/gunicorn wrapper only

(useful for testing)

prerequisites:

• Python3
• Pip

python -m venv virtual_env

virtual_env/bin/pip install -r requirements.txt

linux/mac

source virtual_env/env/activate

windows

./virtual_env/env/activate

and then finally

python main.py

Details

personal website I use including a pytorch image-classifier

Frontend

Simple html/css template with minimal javascript

• Thinking about using a bootstrap theme and rebuild the frontend

Backend

Flask: micro webframework for Python based on Werkzeug, Jinja 2

Gunicorn

Python WSGI HTTP Server for UNIX

acts as webserver accepting requests from Flask

Docker

The Dockerfile builds some layers on top of the pytorch-cpu docker from Deepo

The specific docker is: ufoym/deepo:pytorch-py36-cpu