N.B. I'm sharing this code by request, but I haven't had time yet to clean it up or comment it. In addition, I don't have much experience developing web apps, so this code should simply be seen as a functioning example of a web app providing an interface to a pytorch model, not as an example of best practices. If you find issues with the code, or have advice about best practices that I'm neglecting, please open an issue and let me know.

Dependencies

• python3.6

Installation Instructions:

(this is tested on a desktop running ubuntu 16.04 with a 1080 ti gpu)

• generate secret key and put in line 7 of './flaskr/init.py'
• make virtual env 'python3.6 -m venv ./webdemoVenv36'
• install python packages below
• make symbolic links in /flaskr/static/

ln -s ../../uploaded_ims/ ./
ln -s ../../output_ims/ ./

Python Package Dependencies

• torch 1.5 for cuda 9.2
• imageio
• numpy
• flask
• celery
• waitress

Startup Instructions:

In terminal window/ screen/ tmux session #1:

source webdemoVenv36/bin/activate
waitress-serve --call 'flaskr:create_app'

In terminal window/ screen/ tmux session #2:

source webdemoVenv36/bin/activate
celery -A flaskr worker --loglevel=INFO

The webdemo should now be served on port 8080. If you want to forward port 80 to 8080 (for a nicer web address), see here: https://askubuntu.com/questions/444729/redirect-port-80-to-8080-and-make-it-work-on-local-machine