MonotonicIQA

Table of Contents

• Model Overview
• Requirement
• Getting Started
• Download
• Citation

Model Overview

Requirement

Python 3+
PyTorch 1.4+
MATLAB
Successfully tested on Ubuntu 20.04

Getting Started

Sampling Images

[MATLAB] Sampling images from each datasets

sample_name.m

Mixing Images

[MATLAB] Mixing all the sampled images

combine_pmtrain.m

Train on Mixed Datasets

python Main.py --train True --network basecnn --representation BCNN --batch_size 32 --image_size 384 --lr 3e-4 --decay_interval 3 --decay_ratio 0.9 --max_epochs 24 --backbone resnet34

Get Scores

python Main.py --train False --get_scores True

Result Analysis

calculate_mean.m

Download

Original Paper: Link
Proof of the Transformer: Link

Citation

If you find our work useful in your research, please consider citing it in your publications. We provide a BibTeX entry below.

@article{feng2023learning,
  title     = {Learning from Mixed Datasets: A Monotonic Image Quality Assessment Model},
  author    = {Feng, Zhaopeng and Zhang, Keyang and Jia, Shuyue and Chen, Baoliang and Wang, Shiqi},
  journal   = {IET Electronics Letters},
  volume    = {59},
  number    = {3},
  pages     = {e12698},
  year      = {Jan. 2023},
  publisher = {Wiley Online Library}
}