[Paper] MEON: End-to-End Blind IQA (Image Quality Assessment)

Outperforms IQA-CNN, DeepIQA With Smaller Model Size

Image for post
Image for post

Outline

1. MEON: Network Architecture

Image for post
Image for post
MEON: Network Architecture

1.1. Input and Subtasks

1.2. GDN as Activation Function

Image for post
Image for post

1.3. Shared Layers

1.4. Distortion Type Identification Sub-Network (Sub-Network I)

Image for post
Image for post

1.5. Quality Prediction Sub-Network (Sub-Network II)

Image for post
Image for post
Image for post
Image for post
Image for post
Image for post
Image for post
Image for post

2. MEON: Training and Testing

2.1. Training

Image for post
Image for post
Image for post
Image for post

2.2. Testing

3. Ablation Study

Image for post
Image for post
Median SRCC Across 1000 Sessions on TID2013
Image for post
Image for post
Median SRCC Across 1000 Sessions on TID2013

4. Experimental Results

4.1. Performance on CSIQ and TID2013

Image for post
Image for post
Median SRCC and PLCC results Across 1000 Sessions on CSIQ
Image for post
Image for post
Median SRCC and PLCC results Across 1000 Sessions on TID2013
Image for post
Image for post
D-Test, L-Test, and P-Test on Waterloo Exploration Dataset
Image for post
Image for post
Median SRCC and PLCC results Across 10 Sessions on Full TID2013

4.2. Model Size

Image for post
Image for post
Model Size

Written by

PhD, Researcher. I share what I've learnt and done. :) My LinkedIn: https://www.linkedin.com/in/sh-tsang/, My Paper Reading List: https://bit.ly/33TDhxG

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store