subpic / koniq

KonIQ-10k Deep Learning Models
MIT License
116 stars 22 forks source link
image-quality-assessment

NR-IQA models trained on the KonIQ-10k dataset

This is part of the code for the paper "KonIQ-10k: An ecologically valid database for deep learning of blind image quality assessment". The included Python 2.7 notebooks rely on the kutils library. The Google colab requires the ku library. Project data is available for download from osf.io.

To quickly try out the Koncept512 model:

pip install koncept

Please cite the following paper if you use the code or package:

@article{koniq10k,
author={V. {Hosu} and H. {Lin} and T. {Sziranyi} and D. {Saupe}},
journal={IEEE Transactions on Image Processing},
title={KonIQ-10k: An Ecologically Valid Database for Deep Learning of Blind Image Quality Assessment},
year={2020},
volume={29},
pages={4041-4056}}

Overview

Google colab notebook, Python 3 compatible:

koncept512_train_test_py3_with_kuti.ipynb (updated Sept 2021)

Python 2.7 notebooks:

train_koncept512.ipynb, train_koncept224.ipynb:

train_deeprn.ipynb

metadata/koniq10k_distributions_sets.csv