/image_scoring

Primary LanguagePython

Interview about CV project`

How can we evaluate a picture on its ability to be printable based on a few simple criteria.

  • Face present
  • Smiling people
  • Eyes opened
  • Sharp image
  • If blurry, bokeh effect must be visible
  • Luminous and crisp colors

Application setup

If installed on local, please run

pip install -r requirements.txt

With Django installed, run:

python manage.py runserver

Open your browser to:

http://127.0.0.1:8000/image_score/upload/

Cloud access

This application is running during the process on:

https://image-scoring-build-d26039211250.herokuapp.com/image_score/upload/

Upload a picture and let's take it from there!

Improvements

A lot of improvements as this project stands more as a proof of concept, to use a few feature are indicative on how printable a picture is.

  • Improve the face/smile/eye detection with Data Science projects
  • Improve the use of classic processing of picture to detect sharp images, colored images...
  • The scoring method to mesh all those metrics all together