The Joint Effort Group is an activity of the Video Quality Experts Group (VQEG) created in 2009 that proposes an alternative collaborative action. Usually, VQEG inspects models performance from individual proponents with respects to jointly developed test plans in a competitive manner.
The Joint Effort Group (JEG) is intending to work jointly on both mandatory actions to validate metrics: subjective dataset completion and metrics design. That means that any proposal for improving metrics should be possible to inspect and that it is not necessary to provide a full metric to enter in the process. Therefore, JEG is clearly offering an opportunity to jointly propose quality metrics. In order to maintain fair validation of proposed tool and to avoid precious subjective quality assessment results, the metric has to be designed in a such a way that any contribution is assessed according to the availability of source code to the group. It is also expected to increase after validation for some purposes, to extend the subjective dataset in order to better identify the limitations or extension of application scope of quality metrics. Regarding this latter comment, JEG could produce quality metrics identified through profile/layer with respect of their application scope.
JEG Hybrid is preparing a Call for Proposals and Participation for late 2016. Stay tuned and join the biweekly conference calls!
The following projects are currently ongoing in JEG:
- JEG-Hybrid for H.264/AVC: General framework for objective quality assessment for video using bitstream and decoded video information for H.264/AVC
- JEG-Hybrid database for HEVC: Large-scale database with video bitstreams and results of objective quality measures for HEVC
- JEG-Hybrid characterization of objective measures: Using our large-scale databases to analyze agreement between simple objective measures such as PSNR, SSIM, VIF
- VQEGplayer: 2D and 3D video playback program for performing subjective assessment with remote control and stand-alone facilities
Some more information on JEG and the ongoing project JEG NR hybrid H264 is provided here in the VQEG-JEG:Community_Portal list of this wiki. Discussions concerning single parts of a specific project can be followed on the mailing list.
JEG is organized around shared resources that are used in research projects at the different JEG members.
Consistency of information is judged to be important. Any questions or remarks are therefore welcome.
The following resources are shared by the JEG members and are explained and/or improved on request.
- Databases (SRC, H.264, HEVC, annotation with objective measurements, questions...)
- Organisation of reproducible research: Virtualbox, HMIX files and tools, MySQL processing database
- Robust decoder, packet error introduction and concealment, restreaming
- Processing tool chain
- Statistical analysis of objective Full-Reference measures towards understanding which data needs to be subjectively evaluated
- Contribution from VQEG-MOAVI / Development of indicators for bitstream analysis / indicators for NR measurements
- Combination algorithms for indicators / training / verification / validation including statistical analysis
- Planning of an open call for contributions/competition, updates on Wiki, further plans
- List of Publications
If you are interested in joining JEG, please follow the procedure to join the JEG and JEG-Hybrid mailing list, described here.
Please be aware that VQEG-JEG is a scientific working group. This implies that often results are yet unverified. In many cases, researchers also lack time to document work-in-progress. Please consider this as reason why there is not a lot of written documentation available.
We discuss work in progress on our biweekly conference calls and either officially publish the results or update the documentation when our research is stable and validated.
The best way to get started is:
- Join the VQEG reflector and the VQEG-JEG reflector or contact the co-chairs if you have trouble with this
- Contact the co-chairs, write a mail to the reflector or join a conference call
- Provide a short introduction of yourself and your research on a conference call (5-10 minutes, slides sent to the JEG reflector beforehand are welcome but not mandatory)
- In the following discussion, we will conjointly integrate your research topic and ideas into the currently ongoing JEG research
- More detailed calls or mail exchanges will follow naturally
Participating Groups and Contacts
- Acreo (Sweden)
- AGH University (Poland)
Lucjan Janowski, Mikolaj Leszcuk
- IP-Label.newtest (France)
- Technical University Berlin, T-Labs (Germany)
Marie-Neige Garcia, Werner Robitza
- Ghent University - iMinds (Belgium)
Glenn Van Wallendael
- IRCCyN, University of Nantes (France)
Marcus Barkowsky, Patrick Le Callet, Ahmed Aldahdooh
- Kingston University (United Kingdom)
- University of Vienna (Austria)
- Politecnico di Torino (Italy)
The chairs of the JEG support group are Kjell Brunnström, and Patrick Le Callet.
The chairs of the JEG-Hybrid group are Marcus Barkowsky, Glenn Van Wallendael, and Lucjan Janowski.
The Working Group has biweekly conference calls, announced on the mailing list, (currently on Mondays, 4PM Paris Time CE(S)T). Face–to–face meetings will take place during the biennial VQEG meetings.
The Working Group will utilize the JEG Member mailing list, email@example.com
This text is preliminary and may be incorrect. It is published in order to prevent repetitive discussions on patents.
The JEG group is a cooperative approach. Patents do not necessarily conflict with cooperative approaches.
If a member of JEG wants to file a patent on a particular idea, he cannot discuss it openly, e.g. by using the mailing list, before the idea is filled. He first has to develop the idea, file a patent, get it accepted and then he can discuss it openly and propose it to the JEG for inclusion. When it is known that a proposed algorithm part is patented, this information should be shared within JEG. This applies to both self-owned and third party patents.
In case two or more members of the JEG group wish to have a common patent, the same rules as above apply: They cannot reveal their idea by an open discussion on the mailing list. Practical problems may arise from the fact that patents can usually only be filed by a legal entity and cross-country issues may appear due to different laws.
The purpose of the JEG group is to develop an outstanding objective quality metric. In order to achieve this goal, all algorithm enhancements will be evaluated on an objective basis. This involves, for example, comparing the model performance increase to the computational cost. Patented algorithms will neither be rejected nor preferred.
The JEG group is aware that the (final) developed model will need to be evaluated regarding the patented parts of the algorithm before it can be commercially xploited. It is expected that more than one party is involved. In this phase, a shared solution is advised, e.g. by forming a patent pool. As it is expected that patent owners outside the group need to be contacted, it is welcomed to also have patented parts of the algorithm owned by members of the group in order to facilitate the negotiations. Please note, that this is the last stage of the project and following the current schedule may be expected after 2012.
It was decided on 11th of March 2010 that the topic of patents will only be discussed when there is new relevant information not contained in this text.
Links and further information on VQEG
Edit the Wiki
To modify the wiki's content, a login is required, please contact the co-chairs. Glenn Van Wallendael will assign an account glenn.vanwallendael(at)ugent.be.