Internet Video Coding Exploration

Internet Video Coding Exploration

• 
  Procedure in extending ITM
  
• 
  Assess improvements achieved and the potential for ongoing improvements
  
• 
  Establish plans for future EEs
  

1. General

• 
  Summary
  
  • 
    In this contribution, proponents propose IVC development ground rules: IPR guideline for IVC development, development process, and technology-IPR mapping management.
    
  • 
    The document was reviewed by BoG on IVC with Bog on IVC CfP.
    
• 
  Agreement
  
  • 
    An output doc of this meeting.
    

2. EE

15.0.0.1.1.1.1.1.69m30180 Low-cost multiple-hypothesis motion compensation for internet video coding [Lei Chen, Ronggang Wang, Zhenyu Wang, Siwei Ma, Tiejun Huang, Wen Gao]

• 
  Summary
  
  • 
    In this proposal, proponents recommend a new technology for P frames: low-cost multi-hypothesis motion compensation.
    
  • 
    The experimental results show about 4% BD rate saving in CS2 on average.
    
  • 
    The proponent provides IPR information related to the technology.
    
• 
  Comments
  
  • 
    Q: What is MV1? Is this the same one in MV predictor used in ITM 5.0?
    
    • 
      Yes.
      
• 
  Agreement
  
  • 
    Adopt it to next version of ITM.
    

• 
  Summary
  
  • 
    In this contribution, proponents propose to add 16x8 and 8x16 blocksize in the motion compensation.
    
  • 
    Experimental results show 3.3% and 2.5% bit rate saving on average in CS2 and CS1, respectively
    
  • 
    The proponent provides IPR information related to the technology.
    
• 
  Comments
  
  • 
    There is no the IPR information in the contribution.
    
    • 
      Proponent said that they will include the IPR information to the contribution.
      
• 
  Agreement
  
  • 
    Adopt it to next version of ITM.
    

• 
  Summary
  
  • 
    In this contribution, proponents propose to add a loop filtering tool in the coding loop of Internet Video Coding Test Model.
    
  • 
    Experimental results show that the proposed scheme achieves about 8.9% and 6.0% bit rate saving in CS2, CS1 configure respectively.
    
  • 
    The proponent provides IPR information related to the technology.
    
• 
  There is no the IPR information
  
  • 
    Proponent said that they will include the IPR information to the contribution.
    
• 
  Comments
  
  • 
    Who analyzed the IPR information?
    
    • 
      Proponent said that they investigated by themselves.
      
  • 
    Further study on the IPR information should be done.
    

15.0.0.1.1.1.1.1.72m30183 IVC EE: Performance improvement of ITM5.0 [Ronggang Wang, Lei Chen, Xufeng Li, Hao Lv, Siwei Ma, Tiejun Huang, Wen Gao]

• 
  Summary
  
  • 
    In this contribution, proponents provide information on the performance gain based on the proposed contributions in this meeting: m30180, m30181, and m30182.
    
  • 
    By adding the above three new coding tools in ITM5.0, the coding performance improvement of ITM is 12.1% both in CS2 and 8.4% in CS1.
    
• 
  Comments
  
  • 
    What is the overall gain using m30180 and m30181?
    
    • 
      About 7%, and the proponent will provide the information on the contribution.
      
  • 
    Is the performance on the contribution better than anchor in both constrains?
    
    • 
      Yes.
      
  • 
    Do you think that the performance is comparable compared to anchor?
    
    • 
      Yes.
      
  • 
    The performance in Class D is different from that in other Classes.
    
    • 
      Proponent said that they will investigate the reason why.
      

• 
  Summary
  
  • 
    This is a cross-check report of m30180 on multiple-hypothesis MC.
    
  • 
    The coding efficiency and encoder time of cross-check are identical to the ones reported by proponent.
    

• 
  Summary
  
  • 
    This is a cross-check report of m30181.
    
  • 
    The coding efficiency and encoder time of cross-check are identical to the ones reported by proponent.
    

• 
  Summary
  
  • 
    This is a cross-check report of m30182.
    
  • 
    The coding efficiency and encoder time of cross-check are identical to the ones reported by proponent.
    

• 
  Summary
  
  • 
    The coding efficiency and encoder time of cross-check are identical to the ones reported by proponent.
    

• 
  Summary
  
  • 
    In this contribution, proponents propose to enhance the non-reference P coding in the form of more accurately and reliably detecting which frames should encoded as a non-reference P frame by adopting 2-level non-reference P coding.
    
  • 
    In the experiments, the proposed method gives the average 0.9% bit saving and the coding gain is obtained in the most of the test sequences.
    
  • 
    The proponent provides IPR information related to the technology during this meeting.
    
• 
  Comments
  
  • 
    Do you have IPR information on the technology?
    
    • 
      Further study during the meeting
      
  • 
    What is the two-level approach?
    
    • 
      The current ITM version can make the proposed method similarly but the performance proposed in the contribution is better than the method using current version.
      
    • 
      Signaling method is better than the current version.
      
  • 
    One expert concerns the order of frames because the order of frames on the proposed method seems to make IPR issue.
    
    • 
      There is no IPR issue if the proposed method is used only for P frames.
      
  • 
    Proponent said that they will include the IPR information to the contribution.
    
• 
  Agreement
  
  • 
    Adopt it to next version of ITM, only used in LD case, closed in RA case.
    

• 
  Summary
  
  • 
    This is a cross-check report of m30406.
    
  • 
    The coding efficiency is matched with the ones reported by proponent.
    

• 
  Summary
  
  • 
    Use multiple reference frames to improve the performance of ITM5, -7.3% in CS2 is obtained.
    
  • 
    IPR issue is analyzed on multiple reference frames (MRF), a patent related to MRF filed on 1992, and it has been expired.
    
  • 
    Cross-checked.
    
• 
  Agreement
  
  • 
    Adopt it to next version of ITM only used in LD case, closed in RA case.
    

• 
  Summary
  
  • 
    This is a cross-check report of m30200.
    
  • 
    The coding efficiency is matched with the ones reported by proponent.
    

Topic:

• 
  To adopt the technologies which the group agreed to adopt to the next version of ITM
  
• 
  To further study on IPR information with regard to technologies that are not clear on Type-1 license
  
• 
  To discuss an IVC development ground rule for the future process.
  
• 
  To continue to improve the text as well as the software of next version of ITM.
  
• 
  To discuss moving on to WD stage based on the test results during the meeting.
  

• 
  To adopt the technologies which the group agreed to adopt to the next version of ITM. (done)
  
• 
  To further study on IPR information with regard to technologies that are not clear on Type-1 license.  (done)
  
• 
  To discuss an IVC development ground rule for the future process.  (It will be discussed on Tuesday)
  
• 
  To continue to improve the text as well as the software of next version of ITM.  (Software integration plan: Peking university  Hanyang university  Korea Aerospace University  Peking university), (Text: It is needed to have the editorial period for 4 weeks)
  
• 
  To discuss moving on to WD stage based on the test results during the meeting. (Having time to show the test result based on the adopted technology on Video plenary)
  

• 
  ITM6: Software: Hanyang university (Aug. 3)  Korea Aerospace University (Aug. 10)  Peking university (Aug. 24))
  
• 
  IVC EE: Prof. Wang will prepare the document.
  
• 
  IVC Guidelines: (T.B.D) It will be decided after a discussion on Tuesday.
  

• 
  IVC Guidelines: The related document (i.e., m29775) was reviewed by Bog on IVC with Bog on IVC CfP.
  

Preparation for output documents:

• 
  ITM6
  
  • 
    Text: Hanyang university (Aug. 3)  Korea Aerospace University (Aug. 10)  Peking university (Aug. 24))
    
  • 
    Software: Peking university (Jul. 30)  Hanyang university (Aug. 10)  Korea Aerospace University (Aug. 17)  Peking university (Aug. 24))
    
• 
  IVC EE
  
  • 
    It will be uploaded by Aug. 10.
    
• 
  Development Procedure of Type-1 Standards
  

• 
  It was reported that there are great improvements in compression efficiency over ITM 5.0.
  
• 
  The performance of this temporal version of ITM by adding two tools among the adopted techonogies is comparable with IVC anchor as follows:
  
  • 
    CS1: 9.9% better coding efficiency than IVC anchor
    
  • 
    CS2: 4.3 % worse coding efficiency than IVC anchor
    
• 
  It is expected that the improvement of coding efficiency of ITM 6.0 is higher than the teimporal version for ITM 6.0.
  
• 
  Bog recommends initiating a process to elevate ITM 6.0 to WD stage based on the test result.
  
• 
  Bog recommends initiating a process to manage “Technology-IPR mapping management” document as made decision on the document “Development Procedure of Type-1 Standards”.
  

11. 
    
    Directory: sc29 -> open -> 29view
    29view -> International organisation for standardisation organisation internationale de normalisation
    29view -> International organisation for standardisation organisation internationale de normalisation
    
    
    
    Download 8.47 Mb.
    
    Share with your friends: