/IQA-Distortion-Classification-and-Reconstruction-System

Image Quality Assessment, Image Distortion Classification and Image Reconstruction in a single Django Web system

Primary LanguageJavaScript

IQA, Distortion Classification and Reconstruction System

IQA (Image Quality Assessment)

  • Use multiple models, including CNN, LeNet-5, ResNet, VGG ... to respectively assesses the input image's quality and get the score. The final IQA score of the input image is the average of all these scores.

  • Models are provided in the folder app/assess/models

  • To check the training process or get your own model, follow the link:

    https://github.com/RainFZY/Image-Quality-Assessment-By-Multiple-Models

Image Distortion Classification

  • Classify the distortion of input image into 3 classes:

    noise (wn), blur (gblur), JPEG compression (jpeg)

  • Provide each classification's confidence coefficient.

Image Reconstruction

  • Restore a noise-labeled image to a higher-quality image by noise elimination

Start the Django Service

  1. Enter the project folder in cmd

    cd ../IQA-and-Distortion-Classification-System

  2. run

    python manage.py runserver

  3. Enter http://127.0.0.1:8000/ in your browser

Test Demo

Home page

Test a blurred image

The IQA scores and the distortion classification results are listed in the right area

Test a JPEG compression image

Test a noisy image