VQEG Logo

Video Quality Experts Group (VQEG)

AVHD-AS / P.NATS Phase 2

VEG Audiovisual Model Validation (AVHD-AS) and ITU-T Study Group 12 P.NATS Phase 2

AVHD-AS / P.NATS Phase 2 Project, March 2017 to January 2020 (finished)

AVHD-AS/P.NATS Phase 2 was a joint project of VQEG and ITU-T Study Group 12 Question 14. The goal was to develop a multitude of objective models, varying in terms of complexity / type of input / use-cases for the assessment of video quality in adaptive bitrate streaming services over reliable transport up to 4K.

  • Bitstream models: These models target bitrate ladder derivation, and holistic service and client-side monitoring based on the availability of the relevant bitstream information on the server and client-side respectively.
  • Pixel-based models: Full-reference, reduced-reference, and no-reference models, targeting continuous geographically distributed monitoring and drive testing.
  • Hybrid models: Hybrid no-reference, reduced-reference, and full-reference model, which take both bitstream information and pixels as input.

The project plan specified (step 1) the model validation process and (step 2) a procedure to merge high performing models, to create a single model in each category. In the bitstream model category, high performing models were identified but disagreements in step 2 became an obstacle. High performing models in other categories led to the approval of:

  • ITU-T Rec. P.1204
  • ITU-T Rec. P.1204.3
  • ITU-T Rec. P.1204.4
  • ITU-T Rec. P.1204.5.

The "AVHD-AS / P.NATS Phase II Final Report" is available here.

For a description of the AVHD-AS / P.NATS Phase II validation test, see:

A. Raake et al., "Multi-Model Standard for Bitstream-, Pixel-Based and Hybrid Video Quality Assessment of UHD/4K: ITU-T P.1204," in IEEE Access, vol. 8, pp. 193020-193049, 2020, doi: 10.1109/ACCESS.2020.3032080.

TU Ilmenau created and published a database during this effort. This is not part of the training and validation set. The other datasets are not currently available.

Question relating to this project should be sent to Shahid Mahmood Satti, Silvio Borer, Alexander Raake, and Jörgen Gustafsson.