Deploy water classifier using flask on heroku

A web app to predict whether water is clean or dirty using fastai library.

Demo:

Getting started

• Fork and clone this repo on to your system

Customization

Put your model inside path/models folder

• Place your trained .pth file under path/models/ directory.

Change the class and path name

• Open "app.py" and search for a variable called classes and change that with your own classes

• If your path name is different that "stage-2.pth", then change the learn.load("with_your_file_name.pth") in "app.py"

UI Change

Modify files in templates and static directory.

index.html for the UI and main.js for all the behaviors

Heroku Installation

• Make sure that heroku is installed on your system, if you want to install then follow this https://devcenter.heroku.com/articles/heroku-cli#download-and-install

• Go to https://www.heroku.com/ then login or signup into heroku.

• Once you are logged in, you will have to click on new and click on create a new app as given in the screenshot below

• Give your app name and click on create app

• Go to the folder where you have cloned my repo and follow the steps given on the page

• Once you are done, you will see the link to check your app on browser like this https://water-classifier1.herokuapp.com/

Note: If App is showing Application error then refresh the browser,It will work fine.

Future Roadmap

• Add unit test.
• Add CI/CD integration.
• Create a robust image classifier using the more data or using some advance technique.
• Adding celery and redis as a message brocker to handle the long task.