Task

The task is to develop an Image/Video Super-Resolution Quality Assessment metric.

Motivation

Video and Image Super-resolution (SR) has garnered extensive research in recent years, with new articles appearing monthly. However, as comparison of known image/video quality metrics shows, the task of Super-Resolution Quality Assessment is different from the task of simple Image and Video Quality Assessment (you should compare the leaderboards of our benchmarks: Video Quality Metrics and Super-Resolution Quality Metrics). Therefore, competition of metrics oriented specifically to Super-Resolution is necessary.

Dataset

We provide the participants with train & validation subsets made of Ground-Truth videos and the same videos after applying bicubic downsampling, video codecs and Super-Resolution methods. Extra 15+ hidden test videos will be used for final evaluation. Participants will see test results for 30% of the hidden test videos, full results will be available by the end of the competition.

Dataset will be available as soon as challenge starts.

Organizers from MSU Graphics & Media Lab

Ivan Molodetskikh

Received his M.S. degree in computer science from the Moscow State University in 2020. He is currently a PhD student at the MSU Graphics & Media Lab. His research interests include image inpainting, semantic video matting and machine learning. Ivan had supervised the development of the MSU benchmark of super-resolution for quality enhancement and the development of a super-resolution detection method. Currently he is supervising research related to super-resolution in the Graphics & Media Lab.

Artem Borisov

He is currently in his 3rd year of undergraduate studies. Artem is the main contributor to the MSU Super-Resolution Quality Metrics Benchmark. His research interests include Super-Resolution, Super-Resolution Quality Assessment Metrics and its robustness, as well as Image and Video Quality Assessment Metrics</a>.