quat – video/image quality analysis

quat is a collection of tools, python3 code, and wrappers, to simplify daily quality analysis tasks for video or images. quat is the outcome of several video/image quality models that were developed by Steve Göring.

It consists of several main modules, with specific tasks, e.g. the ml module handles typical machine learning parts, whereas most algorithms and approaches are based on scikit-learn.

Why you should or should not use quat?

quat is a collection of useful parts, it is not a full end-solution or framework that will handle automatically all image/video quality related questions. You can use quat to develop own models, to extract image/video features, and more.

Reference

If you use quat in any research related project, please cite one of the following papers:

@inproceedings{goering2019qomex,
  author={Steve {G{\"o}ring} and Rakesh Rao {Ramachandra Rao} and Alexander Raake},
  title="nofu - A Lightweight {No-Reference} Pixel Based Video Quality Model for
  Gaming Content",
  BOOKTITLE="2019 Eleventh International Conference on Quality of Multimedia Experience
  (QoMEX) (QoMEX 2019)",
  address="Berlin, Germany",
  days=4,
  month=jun,
  year=2019,
  doi={10.1109/QoMEX.2019.8743262},
  ISSN={2472-7814},
  url={https://ieeexplore.ieee.org/document/8743262},
}
@article{goering2021pixel,
  title={Modular Framework and Instances of Pixel-based Video Quality Models for UHD-1/4K},
  author={Steve G\"oring and Rakesh {Rao Ramachandra Rao} and Bernhard Feiten and Alexander Raake},
  journal={IEEE Access},
  volume={},
  pages={},
  year={2021},
  publisher={IEEE},
  note={to appear}
}

nofu is a video quality model using features and general video processing methods that are included in quad.

Contributors

  • Steve Göring

  • Julian Zebelein

  • Serge Molina

  • Rakesh Rao Ramachandra Rao