subjectiveqe-esrgan's Introduction

Subjective Quality Enhancement of Compressed Images using ESRGAN

0. Background

PyTorch implementation of ESRGAN for compressed image subjective quality enhancement.

Feel free to contact: [email protected].

1. Code & Pre-trained Model

[Previous Formal Version]

To unify most of the quality enhancement approaches, we have released the improved ESRGAN at PowerQE.

2. License

We adopt Apache License v2.0. For other licenses, please refer to ESRGAN.

If you find this repository helpful, you may cite:

@incollection{Wang_2019,
	doi = {10.1007/978-3-030-11021-5_5},
	url = {https://doi.org/10.1007%2F978-3-030-11021-5_5},
	year = 2019,
	publisher = {Springer International Publishing},
	pages = {63--79},
	author = {Xintao Wang and Ke Yu and Shixiang Wu and Jinjin Gu and Yihao Liu and Chao Dong and Yu Qiao and Chen Change Loy},
	title = {{ESRGAN}: Enhanced Super-Resolution Generative Adversarial Networks},
	booktitle = {Lecture Notes in Computer Science}
}

@misc{ESRGAN_xing_2020,
  author = {Qunliang Xing},
  title = {Subjective Quality Enhancement of Compressed Images using ESRGAN},
  howpublished = "\url{https://github.com/ryanxingql/subjectiveqe-esrgan}",
  year = {2020}, 
  note = "[Online; accessed 11-April-2021]"
}

subjectiveqe-esrgan's People

Contributors

Stargazers

Watchers

subjectiveqe-esrgan's Issues

Figs in README

Which is the correct YUV version？

Hello, i notice that the png2yuv code in yours is
code = getattr(cv2, 'COLOR_BGR2YCrCb') ycrcb_img = cv2.cvtColor(bgr_img, code).astype(np.uint8) ycrcb_img[:h, :w, 0].tofile(fp) # y ycrcb_img[:h, :w, 2].tofile(fp) # cb (u) ycrcb_img[:h, :w, 1].tofile(fp) # cr (v)

However,the common use YUV in deblock like ARCNN is Matlab version. it is a little differetent with that in opencv
see https://github.com/hkchengrex/PyTorch-ARCNN/blob/master/loader.py

Recommend Projects