List of participants of MSU HDR Video Reconstruction Benchmark: HLG

Name Uses AI Open Source Image/Video restoration Implementation
Maxon No Video
HDRCNN Yes Yes Video*+Image link
HDRTVNet Yes Yes Video link
ExpNet Yes Yes Video*+Image link
DeepHDR Yes Yes Image link
SingleHDR Yes Yes Image link
Akyuz No Yes Image link
Huo No Yes Image link
HuoPhys No Yes Image link
Kuo No Yes Image link
KovOliv No Yes Image link
HDRUNet Yes Yes Image link
KUNet Yes Yes Image link
twostageHDR Yes Yes Image link

* Originally designed for image restoration, but has a video restoration option

Overview of neural network method architectures


UNet-VGG-16 with skip connections neural network.


3 sequentially trained networks:

  • Custom architecture
  • ResNet
  • UNet based GAN


  • Neural network with three parallel branches. They differ in the size of the convolution core
  • Concatenation followed by convolution processing


UNet-VGG-like with skip connections neural network.


Network which is divided into two parts: linearization and restoration.

  • Linearization: custom architecture
  • Restoration: ResNet

10 May 2022
See Also
MSU CVQAD – Compressed VQA Dataset
During our work we have created the database for video quality assessment with subjective scores
Video Upscalers Benchmark
The most extensive comparison of video super-resolution (VSR) algorithms by subjective quality
Site structure