Cleaning Real-World Video Super-Resolution

Home

Discover

Bookmarks

Keep Up With Latest Trending Papers. Computer Science, AI and more.Subscribe

CC BY

Source arXiv

Computer Vision

Cleaning Real-World Video Super-Resolution

Investigating Tradeoffs in Real-World Video Super-Resolution

We propose a stochastic degradation scheme that reduces up to 40\% of training time without sacrificing performance.

We then analyze different training settings and suggest that employing longer sequences rather than larger batches during training allows more effective uses of temporal information, leading to more stable performance during inference.

To facilitate fair comparisons, we propose the new videolq dataset, which contains a large variety of real-world low-quality video sequences containing rich textures and patterns.

Our dataset can serve as a common ground for benchmarking.

Authors

Kelvin C.K. Chan, Shangchen Zhou, Xiangyu Xu, Chen Change Loy

Related Topics

Propagation

Image pre-cleaning stage

Cleaning module

Realbasicvsr

Real-world vsr models

Long-term propagation

Stochastic degradation scheme

Notes

As a result, it generatesmuch more details in fine regions, improving visual qual-ity. For instance, only RealBasicVSR is able to recover theword “hobby”, which can be clearly seen in other frames.

Andy P

Like Tech...

Impressive results!

Read the Paper

Read Paper

Home

Discover

Bookmarks

◐ Recommended Follows

◐ Latest Activity