Improving Open-Source HEVC Encoding
Comcast, Intel, MulticoreWare and Ozyegin University
- Alex Giladi, Comcast
- Sravanthi Kota Venkata, Intel
- Pradeep Ramachandran, MulticoreWare
- Ali C. Begen, Ozyegin University
Video streaming traffic accounts for a significant share of the Internet traffic. Per the latest Sandvine report, the share of downstream video streaming traffic reached 58%. A significant portion of this traffic is professionally produced linear and cinematic content. The amount of available premium UltraHD content is growing fast, with hundreds of movies produced or remastered, and a growing number of popular events and linear channels produced in UltraHD. For example, the final game of the World Cup 2018 was produced in UltraHD and (according to Conviva report) had 9.8 million concurrent streams.
High quality and efficient compression of this content is a challenge for the industry due to the high bitrates required for its transmission. HEVC is the codec of choice for UltraHD content, and the open source x265 project is currently the most widely used HEVC encoder. The goal of this challenge is to produce significant algorithmic improvements to HEVC encoding, and make them available to the open source community.
- Software: x265, latest version
The entry requirements for the competition are:
- Making the proposed algorithmic improvement available as a patch to the most recent version of x265. The patch must be made available via a publicly accessible URL, and must be licensed as GPLv2. Patch acceptance and merge into x265 are not required.
- Independence of the proposed improvement from special purpose hardware or proprietary software. Dependence on open source software such as OpenCV or TensorFlow is permitted.
- Compliance with Main/Main10 profile and level of 5.1 or lower.
Each entry will be judged according to the following criteria:
- Significance will be assessed using the VMAF quality metric and a set of several hours of professionally produced UltraHD and HD content.
- The content will not be made available to the contestants. The improvement can apply to a subset of content (e.g., only HDR or only 8 or 10 bits, only animation, etc.) and will be measured using the "veryslow" preset of x265 in VBV and VBV+crf modes, in both psnr and ssim tune modes.
- Run-time computational complexity of the proposed algorithm, assuming Intel Sky Lake-generation or newer Xeon CPU for assessing the potential of assembly optimization of the proposed technique. Assembly optimization is not expected.
- Submission (for all entries): May 27th, 2019
- Acceptance notification (for entries with a companion paper): July 1st, 2019
- Camera-ready submission (for entries with a companion paper) : July 15th, 2019
- Winners announcement (for all entries): Aug. 1st, 2019
- Entry presentations (for all entries): During the ICIP week
Submissions should provide enough details for the implemented software patch. A technical paper along with the patch code is preferred, but not required, by the submission deadline indicated above.
The authors of the companion papers will be notified after a technical review process and the authors of the accepted papers need to prepare a camera-ready version so that their papers can be published by IEEEXplore under the name of "International Conference on Image Processing Challenges (ICIPC)". For the exact formatting guidelines (e.g., template, page limit, etc.), refer to the ICIP 2019 Website.
Winners are to be chosen by a committee appointed by the sponsoring organizations and results will be final, which will be announced by Aug. 1st. If contributions of sufficient quality are not received, then some or all of the awards may not be granted. The challenge is open to any individual, commercial or academic institution, however, submissions from the sponsoring organizations are not eligible for the awards.
Submission site will open soon.
The winner will be awarded the prize of 6000 USD, and the runner-up will be awarded 3000 USD.
Questions and Requests
If you have any questions or requests, or need further clarifications, please contact the organizers.