Applied Filters
- Dmitriy Vatolin
- AuthorRemove filter
People
Colleagues
- Anastasia Antsiferova (3)
- Dmitriy L Kulikov (3)
- Mikhail Erofeev (2)
- Sergey V Grishin (2)
- A Moskalenko (1)
- A V Zvezdakova (1)
- Alexander Gushchin (1)
- Alexander Ploshkin (1)
- Dmitriy Popov (1)
- Ekaterina Shumitskaya (1)
- Karen Simonyan (1)
- Maksim Siniukov (1)
- Maksim Smirnov (1)
- Mikhail Voronin (1)
- Nickolay Safonov (1)
- S V Zvezdakov (1)
- Sergey Lavrushkin (1)
- Vitaliy Lyudvichenko (1)
Publication
Proceedings/Book Names
- AICCC '21: Proceedings of the 2021 4th Artificial Intelligence and Cloud Computing Conference (1)
- ICDSP '24: Proceedings of the 2024 8th International Conference on Digital Signal Processing (1)
- IMIP '19: Proceedings of the 2019 International Conference on Intelligent Medicine and Image Processing (1)
- NIPS '22: Proceedings of the 36th International Conference on Neural Information Processing Systems (1)
- SCCG '08: Proceedings of the 24th Spring Conference on Computer Graphics (1)
Publication Date
Export Citations
Publications
Save this search
Please login to be able to save your searches and receive alerts for new content matching your search criteria.
- research-articlePublished By ACMPublished By ACM
A New HDR Video Reconstruction Benchmark, Dataset and Metric
- Mikhail Voronin
Lomonosov Moscow State University, Russia
, - Nickolay Safonov
Lomonosov Moscow State University, Russia
, - Dmitriy Vatolin
Lomonosov Moscow State University, Russia and MSU Institute for Artificial Intelligence, Russian Federation
ICDSP '24: Proceedings of the 2024 8th International Conference on Digital Signal Processing•February 2024, pp 27-32• https://doi.org/10.1145/3653876.3653896This paper discusses the challenges of evaluating SDR-to-HDR methods. The lack of comprehensive comparisons between existing methods prompted us to independently evaluate 14 HDR reconstruction approaches. To achieve this, we created a new closed HDR ...
- 0Citation
- 17
- Downloads
MetricsTotal Citations0Total Downloads17Last 12 Months17Last 6 weeks2
- Mikhail Voronin
- research-article
Towards adversarial robustness verification of no-reference image- and video-quality metrics
- Ekaterina Shumitskaya
Graphics and Media Lab, Lomonosov Moscow State University, Moscow, 119992, Russian Federation
, - Anastasia Antsiferova
Institute for Artificial Intelligence, Lomonosov Moscow State University, Moscow, 119192, Russian Federation
ISP RAS Research Center for Trusted Artificial Intelligence, Ivannikov Institute for System Programming of the RAS, Moscow, 109004, Russian Federation
, - Dmitriy Vatolin
Graphics and Media Lab, Lomonosov Moscow State University, Moscow, 119992, Russian Federation
ISP RAS Research Center for Trusted Artificial Intelligence, Ivannikov Institute for System Programming of the RAS, Moscow, 109004, Russian Federation
Institute for Artificial Intelligence, Lomonosov Moscow State University, Moscow, 119192, Russian Federation
Computer Vision and Image Understanding, Volume 240, Issue C•Mar 2024 • https://doi.org/10.1016/j.cviu.2023.103913AbstractIn this paper, we propose a new method of analysing the stability of modern deep image- and video-quality metrics to different adversarial attacks. The stability analysis of quality metrics is becoming important because nowadays the majority of ...
Highlights- Modern image- and video-quality metrics are vulnerable to adversarial attacks.
- It is possible to attack metrics fast even for high-resolution images/video frames.
- Metrics that currently lead in existing benchmarks are most ...
- 0Citation
MetricsTotal Citations0
- Ekaterina Shumitskaya
- research-article
Video compression dataset and benchmark of learning-based video-quality metrics
- Anastasia Antsiferova
ISP RAS Research Center for Trusted Artificial Intelligence and MSU Institute for Artificial Intelligence
, - Sergey Lavrushkin
ISP RAS Research Center for Trusted Artificial Intelligence and MSU Institute for Artificial Intelligence
, - Maksim Smirnov
Lomonosov Moscow State University
, - Alexander Gushchin
Lomonosov Moscow State University
, - Dmitriy Vatolin
ISP RAS Research Center for Trusted Artificial Intelligence and MSU Institute for Artificial Intelligence and Lomonosov Moscow State University
, - Dmitriy Kulikov
MSU Institute for Artificial Intelligence and Lomonosov Moscow State University
NIPS '22: Proceedings of the 36th International Conference on Neural Information Processing Systems•November 2022, Article No.: 1004, pp 13814-13825Video-quality measurement is a critical task in video processing. Nowadays, many implementations of new encoding standards — such as AV1, VVC, and LCEVC — use deep-learning-based decoding algorithms with perceptual metrics that serve as optimization ...
- 0Citation
MetricsTotal Citations0- 1
Supplementary Material3600270.3601274_supp.pdf
- Anastasia Antsiferova
- research-articlePublished By ACMPublished By ACM
Hacking VMAF and VMAF NEG: Vulnerability to Different Preprocessing Methods
- Maksim Siniukov
Lomonosov Moscow State University, Russia
, - Anastasia Antsiferova
Lomonosov Moscow State University, Russia
, - Dmitriy Kulikov
Lomonosov Moscow State University, Russia
, - Dmitriy Vatolin
Lomonosov Moscow State University, Russia
AICCC '21: Proceedings of the 2021 4th Artificial Intelligence and Cloud Computing Conference•December 2021, pp 89-96• https://doi.org/10.1145/3508259.3508272Video quality measurement plays a critical role in the development of video processing applications. In this paper, we show how popular quality metrics VMAF and its tuning-resistant version VMAF NEG can be artificially increased by video preprocessing. ...
- 8Citation
- 87
- Downloads
MetricsTotal Citations8Total Downloads87Last 12 Months17Last 6 weeks3
- Maksim Siniukov
- research-article
Method for Enhancing High-Resolution Image Inpainting with Two-Stage Approach
- A. Moskalenko
Lomonosov Moscow State University, Moscow, Russia, Faculty of Computational Mathematics and Cybernetics, 119991, Moscow, Russia
, - M. Erofeev
Lomonosov Moscow State University, Moscow, Russia, Faculty of Computational Mathematics and Cybernetics, 119991, Moscow, Russia
, - D. Vatolin
Lomonosov Moscow State University, Moscow, Russia, Faculty of Computational Mathematics and Cybernetics, 119991, Moscow, Russia
Programming and Computing Software, Volume 47, Issue 3•May 2021, pp 201-206 • https://doi.org/10.1134/S0361768821030075AbstractIn recent years, the field of image inpainting has developed rapidly, learning based approaches show impressive results in the task of filling missing parts in an image. But most deep methods are strongly tied to the resolution of the images on ...
- 0Citation
MetricsTotal Citations0
- A. Moskalenko
- research-article
BSQ-rate: a New Approach for Video-codec Performance Comparison and Drawbacks of Current Solutions
- A. V. Zvezdakova
Moscow State University, Moscow, Russia
, - D. L. Kulikov
Moscow State University, Moscow, Russia
Dubna State University, Dubna, Russia
, - S. V. Zvezdakov
Moscow State University, Moscow, Russia
, - D. S. Vatolin
Moscow State University, Moscow, Russia
Programming and Computing Software, Volume 46, Issue 3•May 2020, pp 183-194 • https://doi.org/10.1134/S0361768820030111AbstractThis paper is dedicated to the analysis of the existing approaches to video codecs comparisons. It includes the revealed drawbacks of popular comparison methods and proposes new techniques. The performed analysis of user-generated videos ...
- 2Citation
MetricsTotal Citations2
- A. V. Zvezdakova
- research-articlePublished By ACMPublished By ACM
Improving Video Compression with Deep Visual-attention Models
- Vitaliy Lyudvichenko
Lomonosov Moscow State University, Russia
, - Mikhail Erofeev
Lomonosov Moscow State University, Russia
, - Alexander Ploshkin
Lomonosov Moscow State University, Russia
, - Dmitriy Vatolin
Lomonosov Moscow State University, Russia
IMIP '19: Proceedings of the 2019 International Conference on Intelligent Medicine and Image Processing•April 2019, pp 88-94• https://doi.org/10.1145/3332340.3332358Recent advances in deep learning have markedly improved the quality of visual-attention modelling. In this work we apply these advances to video compression. We propose a compression method that uses a saliency model to adaptively compress frame areas ...
- 7Citation
- 173
- Downloads
MetricsTotal Citations7Total Downloads173Last 12 Months25
- Vitaliy Lyudvichenko
- article
Double up-conversion of video frame rate based on bidirectional motion compensation
- D. S. Vatolin
Graphics and Multimedia Laboratory, Department of Computational Mathematics and Cybernetics, Moscow State University, Moscow, Russia 119992
, - S. V. Grishin
Graphics and Multimedia Laboratory, Department of Computational Mathematics and Cybernetics, Moscow State University, Moscow, Russia 119992
Programming and Computing Software, Volume 35, Issue 6•November 2009, pp 351-364 • https://doi.org/10.1134/S036176880906005XAn algorithm for video sequence frame rate double up-conversion is described. The algorithm is based on bidirectional motion compensation. The algorithm has no smoothing in the time domain implying the absence of the image sharpness oscillations in the ...
- 1Citation
MetricsTotal Citations1
- D. S. Vatolin
- research-articlePublished By ACMPublished By ACM
Video super-resolution using motion compensation and classification-aided fusion
- Karen Simonyan
Moscow State University
, - Sergey Grishin
Moscow State University
, - Dmitriy Vatolin
YUVsoft Corp.
, - Dmitriy Popov
YUVsoft Corp.
SCCG '08: Proceedings of the 24th Spring Conference on Computer Graphics•April 2008, pp 143-148• https://doi.org/10.1145/1921264.1921294In this paper, we propose a super-resolution algorithm based on image fusion via pixel classification. Two high-resolution images are constructed, the first by means of motion compensation and the second by means of image interpolation. The AdaBoost ...
- 1Citation
- 125
- Downloads
MetricsTotal Citations1Total Downloads125Last 12 Months1Last 6 weeks1
- Karen Simonyan
Author Profile Pages
- Description: The Author Profile Page initially collects all the professional information known about authors from the publications record as known by the ACM bibliographic database, the Guide. Coverage of ACM publications is comprehensive from the 1950's. Coverage of other publishers generally starts in the mid 1980's. The Author Profile Page supplies a quick snapshot of an author's contribution to the field and some rudimentary measures of influence upon it. Over time, the contents of the Author Profile page may expand at the direction of the community.
Please see the following 2007 Turing Award winners' profiles as examples: - History: Disambiguation of author names is of course required for precise identification of all the works, and only those works, by a unique individual. Of equal importance to ACM, author name normalization is also one critical prerequisite to building accurate citation and download statistics. For the past several years, ACM has worked to normalize author names, expand reference capture, and gather detailed usage statistics, all intended to provide the community with a robust set of publication metrics. The Author Profile Pages reveal the first result of these efforts.
- Normalization: ACM uses normalization algorithms to weigh several types of evidence for merging and splitting names.
These include:- co-authors: if we have two names and cannot disambiguate them based on name alone, then we see if they have a co-author in common. If so, this weighs towards the two names being the same person.
- affiliations: names in common with same affiliation weighs toward the two names being the same person.
- publication title: names in common whose works are published in same journal weighs toward the two names being the same person.
- keywords: names in common whose works address the same subject matter as determined from title and keywords, weigh toward being the same person.
The more conservative the merging algorithms, the more bits of evidence are required before a merge is made, resulting in greater precision but lower recall of works for a given Author Profile. Many bibliographic records have only author initials. Many names lack affiliations. With very common family names, typical in Asia, more liberal algorithms result in mistaken merges.
Automatic normalization of author names is not exact. Hence it is clear that manual intervention based on human knowledge is required to perfect algorithmic results. ACM is meeting this challenge, continuing to work to improve the automated merges by tweaking the weighting of the evidence in light of experience.
- Bibliometrics: In 1926, Alfred Lotka formulated his power law (known as Lotka's Law) describing the frequency of publication by authors in a given field. According to this bibliometric law of scientific productivity, only a very small percentage (~6%) of authors in a field will produce more than 10 articles while the majority (perhaps 60%) will have but a single article published. With ACM's first cut at author name normalization in place, the distribution of our authors with 1, 2, 3..n publications does not match Lotka's Law precisely, but neither is the distribution curve far off. For a definition of ACM's first set of publication statistics, see Bibliometrics
- Future Direction:
The initial release of the Author Edit Screen is open to anyone in the community with an ACM account, but it is limited to personal information. An author's photograph, a Home Page URL, and an email may be added, deleted or edited. Changes are reviewed before they are made available on the live site.
ACM will expand this edit facility to accommodate more types of data and facilitate ease of community participation with appropriate safeguards. In particular, authors or members of the community will be able to indicate works in their profile that do not belong there and merge others that do belong but are currently missing.
A direct search interface for Author Profiles will be built.
An institutional view of works emerging from their faculty and researchers will be provided along with a relevant set of metrics.
It is possible, too, that the Author Profile page may evolve to allow interested authors to upload unpublished professional materials to an area available for search and free educational use, but distinct from the ACM Digital Library proper. It is hard to predict what shape such an area for user-generated content may take, but it carries interesting potential for input from the community.
Bibliometrics
The ACM DL is a comprehensive repository of publications from the entire field of computing.
It is ACM's intention to make the derivation of any publication statistics it generates clear to the user.
- Average citations per article = The total Citation Count divided by the total Publication Count.
- Citation Count = cumulative total number of times all authored works by this author were cited by other works within ACM's bibliographic database. Almost all reference lists in articles published by ACM have been captured. References lists from other publishers are less well-represented in the database. Unresolved references are not included in the Citation Count. The Citation Count is citations TO any type of work, but the references counted are only FROM journal and proceedings articles. Reference lists from books, dissertations, and technical reports have not generally been captured in the database. (Citation Counts for individual works are displayed with the individual record listed on the Author Page.)
- Publication Count = all works of any genre within the universe of ACM's bibliographic database of computing literature of which this person was an author. Works where the person has role as editor, advisor, chair, etc. are listed on the page but are not part of the Publication Count.
- Publication Years = the span from the earliest year of publication on a work by this author to the most recent year of publication of a work by this author captured within the ACM bibliographic database of computing literature (The ACM Guide to Computing Literature, also known as "the Guide".
- Available for download = the total number of works by this author whose full texts may be downloaded from an ACM full-text article server. Downloads from external full-text sources linked to from within the ACM bibliographic space are not counted as 'available for download'.
- Average downloads per article = The total number of cumulative downloads divided by the number of articles (including multimedia objects) available for download from ACM's servers.
- Downloads (cumulative) = The cumulative number of times all works by this author have been downloaded from an ACM full-text article server since the downloads were first counted in May 2003. The counts displayed are updated monthly and are therefore 0-31 days behind the current date. Robotic activity is scrubbed from the download statistics.
- Downloads (12 months) = The cumulative number of times all works by this author have been downloaded from an ACM full-text article server over the last 12-month period for which statistics are available. The counts displayed are usually 1-2 weeks behind the current date. (12-month download counts for individual works are displayed with the individual record.)
- Downloads (6 weeks) = The cumulative number of times all works by this author have been downloaded from an ACM full-text article server over the last 6-week period for which statistics are available. The counts displayed are usually 1-2 weeks behind the current date. (6-week download counts for individual works are displayed with the individual record.)
ACM Author-Izer Service
Summary Description
ACM Author-Izer is a unique service that enables ACM authors to generate and post links on both their homepage and institutional repository for visitors to download the definitive version of their articles from the ACM Digital Library at no charge.
Downloads from these sites are captured in official ACM statistics, improving the accuracy of usage and impact measurements. Consistently linking to definitive version of ACM articles should reduce user confusion over article versioning.
ACM Author-Izer also extends ACM’s reputation as an innovative “Green Path” publisher, making ACM one of the first publishers of scholarly works to offer this model to its authors.
To access ACM Author-Izer, authors need to establish a free ACM web account. Should authors change institutions or sites, they can utilize the new ACM service to disable old links and re-authorize new links for free downloads from a different site.
How ACM Author-Izer Works
Authors may post ACM Author-Izer links in their own bibliographies maintained on their website and their own institution’s repository. The links take visitors to your page directly to the definitive version of individual articles inside the ACM Digital Library to download these articles for free.
The Service can be applied to all the articles you have ever published with ACM.
Depending on your previous activities within the ACM DL, you may need to take up to three steps to use ACM Author-Izer.
For authors who do not have a free ACM Web Account:
- Go to the ACM DL http://dl.acm.org/ and click SIGN UP. Once your account is established, proceed to next step.
For authors who have an ACM web account, but have not edited their ACM Author Profile page:
- Sign in to your ACM web account and go to your Author Profile page. Click "Add personal information" and add photograph, homepage address, etc. Click ADD AUTHOR INFORMATION to submit change. Once you receive email notification that your changes were accepted, you may utilize ACM Author-izer.
For authors who have an account and have already edited their Profile Page:
- Sign in to your ACM web account, go to your Author Profile page in the Digital Library, look for the ACM Author-izer link below each ACM published article, and begin the authorization process. If you have published many ACM articles, you may find a batch Authorization process useful. It is labeled: "Export as: ACM Author-Izer Service"
ACM Author-Izer also provides code snippets for authors to display download and citation statistics for each “authorized” article on their personal pages. Downloads from these pages are captured in official ACM statistics, improving the accuracy of usage and impact measurements. Consistently linking to the definitive version of ACM articles should reduce user confusion over article versioning.
Note: You still retain the right to post your author-prepared preprint versions on your home pages and in your institutional repositories with DOI pointers to the definitive version permanently maintained in the ACM Digital Library. But any download of your preprint versions will not be counted in ACM usage statistics. If you use these AUTHOR-IZER links instead, usage by visitors to your page will be recorded in the ACM Digital Library and displayed on your page.
FAQ
- Q. What is ACM Author-Izer?
A. ACM Author-Izer is a unique, link-based, self-archiving service that enables ACM authors to generate and post links on either their home page or institutional repository for visitors to download the definitive version of their articles for free.
- Q. What articles are eligible for ACM Author-Izer?
- A. ACM Author-Izer can be applied to all the articles authors have ever published with ACM. It is also available to authors who will have articles published in ACM publications in the future.
- Q. Are there any restrictions on authors to use this service?
- A. No. An author does not need to subscribe to the ACM Digital Library nor even be a member of ACM.
- Q. What are the requirements to use this service?
- A. To access ACM Author-Izer, authors need to have a free ACM web account, must have an ACM Author Profile page in the Digital Library, and must take ownership of their Author Profile page.
- Q. What is an ACM Author Profile Page?
- A. The Author Profile Page initially collects all the professional information known about authors from the publications record as known by the ACM Digital Library. The Author Profile Page supplies a quick snapshot of an author's contribution to the field and some rudimentary measures of influence upon it. Over time, the contents of the Author Profile page may expand at the direction of the community. Please visit the ACM Author Profile documentation page for more background information on these pages.
- Q. How do I find my Author Profile page and take ownership?
- A. You will need to take the following steps:
- Create a free ACM Web Account
- Sign-In to the ACM Digital Library
- Find your Author Profile Page by searching the ACM Digital Library for your name
- Find the result you authored (where your author name is a clickable link)
- Click on your name to go to the Author Profile Page
- Click the "Add Personal Information" link on the Author Profile Page
- Wait for ACM review and approval; generally less than 24 hours
- Q. Why does my photo not appear?
- A. Make sure that the image you submit is in .jpg or .gif format and that the file name does not contain special characters
- Q. What if I cannot find the Add Personal Information function on my author page?
- A. The ACM account linked to your profile page is different than the one you are logged into. Please logout and login to the account associated with your Author Profile Page.
- Q. What happens if an author changes the location of his bibliography or moves to a new institution?
- A. Should authors change institutions or sites, they can utilize ACM Author-Izer to disable old links and re-authorize new links for free downloads from a new location.
- Q. What happens if an author provides a URL that redirects to the author’s personal bibliography page?
- A. The service will not provide a free download from the ACM Digital Library. Instead the person who uses that link will simply go to the Citation Page for that article in the ACM Digital Library where the article may be accessed under the usual subscription rules.
However, if the author provides the target page URL, any link that redirects to that target page will enable a free download from the Service.
- Q. What happens if the author’s bibliography lives on a page with several aliases?
- A. Only one alias will work, whichever one is registered as the page containing the author’s bibliography. ACM has no technical solution to this problem at this time.
- Q. Why should authors use ACM Author-Izer?
- A. ACM Author-Izer lets visitors to authors’ personal home pages download articles for no charge from the ACM Digital Library. It allows authors to dynamically display real-time download and citation statistics for each “authorized” article on their personal site.
- Q. Does ACM Author-Izer provide benefits for authors?
- A. Downloads of definitive articles via Author-Izer links on the authors’ personal web page are captured in official ACM statistics to more accurately reflect usage and impact measurements.
Authors who do not use ACM Author-Izer links will not have downloads from their local, personal bibliographies counted. They do, however, retain the existing right to post author-prepared preprint versions on their home pages or institutional repositories with DOI pointers to the definitive version permanently maintained in the ACM Digital Library.
- Q. How does ACM Author-Izer benefit the computing community?
- A. ACM Author-Izer expands the visibility and dissemination of the definitive version of ACM articles. It is based on ACM’s strong belief that the computing community should have the widest possible access to the definitive versions of scholarly literature. By linking authors’ personal bibliography with the ACM Digital Library, user confusion over article versioning should be reduced over time.
In making ACM Author-Izer a free service to both authors and visitors to their websites, ACM is emphasizing its continuing commitment to the interests of its authors and to the computing community in ways that are consistent with its existing subscription-based access model.
- Q. Why can’t I find my most recent publication in my ACM Author Profile Page?
- A. There is a time delay between publication and the process which associates that publication with an Author Profile Page. Right now, that process usually takes 4-8 weeks.
- Q. How does ACM Author-Izer expand ACM’s “Green Path” Access Policies?
- A. ACM Author-Izer extends the rights and permissions that authors retain even after copyright transfer to ACM, which has been among the “greenest” publishers. ACM enables its author community to retain a wide range of rights related to copyright and reuse of materials. They include:
- Posting rights that ensure free access to their work outside the ACM Digital Library and print publications
- Rights to reuse any portion of their work in new works that they may create
- Copyright to artistic images in ACM’s graphics-oriented publications that authors may want to exploit in commercial contexts
- All patent rights, which remain with the original owner