About Us

We are the part of the Computer Graphics and Multimedia Laboratory, situated in the Moscow State University, Russia. We are:

• #1 university in Russia, by THE, QS and ARWU rankings
• #1 computer graphics and multimedia laboratory in Russia
• In the top of the world university laboratories for the research applicability and video compression:
  • #1 codecs comparison
  • #1 video-matting benchmark
  • #1 video quality measurement tool
  • #1 subjective quality estimation
  • #1 by the number of implemented stereo metrics

The case is that the governmental financial support is considerable low, in compare to other countries. Hence we have the only way to survive - to be extremely practical.

Here is a list of companies we collaborated with:

Graphics & Media Lab Video Group is part of Lomonosov Moscow State University Computer Science Department. The history of MSU’s Video Group began at the end of the 1980s. Main research avenues include areas of computer graphics, computer vision, and video processing. Video Group works in areas of video compression (codecs testing and tuning, quality metrics research and development), video and S3D video quality analysis and benchmarking. We developed exclusive tools for Intel, Samsung, Huawei, RealNetworks, and other companies, adapting our algorithms for specific video streams, applications, and hardware like TV sets, graphics cards, etc.

Main research areas

Video Codecs Analysis

Worldwide leading MSU Video Codecs Comparisons: conducted annually since 2007

• Collaborations with: Google, Intel, AMD, NVIDIA, Huawei, Tencent, Alibaba, and others
• Detailed reports: 60+ Full HD and 4K videos, 8-10 different target bitrates, and 15000+ resulting graphs
• Accurate subjective analysis powered by our Subjectify.us platform
• 27+ of our reports with more than 400.000 downloads are publically available

Our content-aware codec tuning service improves encoding performance on a wide range of videos and encoding use cases, reducing bitrate up to 40% and encoding time up to 50% with the same quality

Video Quality Estimation

MSU VQMT is a tool for performing video/image quality analyses using reference or no-reference metrics

• Widest range of metrics and formats, including 20+ objective metrics
• GPU support: up to 11.7x speedup with GPU
• Fastest implementation of VMAF
• Fastest SSIM/MS-SSIM speed on 4K/8K video

VQMT3D is a tool for stereoscopic video quality estimation

• Predict in advance whether you will have a headache after watching the 3D movie
• World’s largest set of 18 metrics, including 5 unique ones, provide quality estimation of 2D-to-3D conversion
• Over 100 analyzed 3D Blu-ray movies and 1,000 VR180 videos with 12 reports
• The largest study of stereoscopic error influence on 300+ viewers

Subjective Quality Estimation

Subjectify.us is a crowd-sourced evaluation platform designed for subjective comparison of video, images, and sound processing methods

• 10x cut the budget of your subjective study
• Case studies:
  • Video codecs comparison with 6,100+ participants
  • Comparison of video completion methods with 341 participants
  • Netflix study replication with 375 participants

Benchmarking

We launched 6 different benchmarks and expect to launch 14 more by the end of 2022

• Video Matting Benchmark: the first public objective benchmark for video-matting methods
• Video Super-Resolution Benchmark, which has the biggest amount of participants (20+ methods) on paperswithcode
• Deinterlacing Benchmark: The first benchmark for comparison of deinterlacing methods
• Mobile Video Codecs Benchmark: The benchmark estimates key indicators (power efficiency and speed) to find the best mobile video-decoder on a given Android device
• Shot Boundary Detection Benchmark: Comprehensive analysis of shot boundary detection methods on extensive and diverse datasets
• Super-Resolution for Video Compression Benchmark: The comparison between super-resolution models on videos, compressed with different codecs

Other

• Our Compression project is the biggest resource on video compression.
• We make algorithms to correct and improve videos. Our video filters have over 3 million downloads.

Contacts

90% of our projects are sponsored by companies. We have experience of a long-term collaboration with Intel, Samsung, Huawei, and others. All our research is aimed to be extremely practical for the industry.

We are open to cooperation in the fields of video processing, video compression, and video quality analysis.

Feel free to contact us: dmitriy@graphics.cs.msu.ru