by dmaniry

dmaniry /deepIQA

NR and FR IQA models based on Deep Convolutional Neural Networks

138 Stars 63 Forks Last release: Not found MIT License 14 Commits 0 Releases

Available items

No Items, yet!

The developer of this repository has not created any items for sale yet. Need a bug fixed? Help with integration? A different license? Create a request here:


This is the reference implementation of Deep Neural Networks for No-Reference and Full-Reference Image Quality Assessment. The pretrained models contained in the models directory were trained for both NR and FR IQA and for both model variants described in the paper. They were trained on the full LIVE or TID2013 database respectively, as used in the cross-dataset evaluations. This evaluation script uses non-overlapping 32x32 patches to produce deterministic scores, whereas the evaluation in the paper uses randomly sampled overlapping patches.

usage: [-h] [--model MODEL] [--top {patchwise,weighted}] [--gpu GPU] INPUT [REF]


  • chainer
  • ~~scikit-learn~~
  • ~~opencv~~


  • add training code
  • add cpu support (minor change)
  • ~~remove opencv and scikit-learn dependencies for loading data (minor changes)~~
  • ~~fix non-deterministic behaviour~~

We use cookies. If you continue to browse the site, you agree to the use of cookies. For more information on our use of cookies please see our Privacy Policy.