Search Results for "JVET-AG0065"
Found 6 document(s)
Search documents
Use number, keyword, author, or MPEG number. Filter by meeting when needed.
Not signed in yet.
Abstract
This contribution proposes a method of using adaptive coefficient precision for CCALF to replace the fixed 7-bit precision currently used in ECM. The proposed method is implemented on top of ECM-11.0 software, and its performance is reported as follows:
Loading full meeting notes...
JVET-AG0065 Non-EE2: Adaptive coefficient precision for CCALF [N. Song, Y. Yu, H. Yu, D. Wang (OPPO)]
This contribution proposes a method of using adaptive coefficient precision for CCALF to replace the fixed 7-bit precision currently used in ECM. The proposed method is implemented on top of ECM-11.0 software, and its performance is reported as follows:
AI: { 0.00% Y, -0.20% U, -0.25% V, 100.2% EncT, 100.5% DecT }
RA: { 0.00% Y, -0.26% U, -0.39% V, 99.9% EncT, 100.0% DecT }
EE2-4.3 which proposes to use adaptive coefficient precision for luma ALF was adopted in this meeting. The proponent of EE2-4.3 reported that adaptive coefficient precision for chroma ALF was attempted but brought no further gain.
This contribution is closely related to JVET-AG0233, which also proposes to apply adaptive coefficient precision to CCALF, and was already agreed to be investigated in the next round of EE.
The set of precisions in this contribution {7, 8, 9, 10} is the same as proposed in JVET-AG0233. Both propopents agreed to test the two contributions (JVET-AG0233 and JVET-AG0065) as one EE test set.
It was commented that the set of precisions for luma ALF is {5,6,7,8}, and it would be interesting to test this set on CCALF as well. This was agreed.
It was agreed to investigate this in the EE together with JVET-AG0233.
Tests to be conducted should include:
- Study the effect of coefficient constraint removal and adaptive precision separately for CCALF
- Study using the set of precisions {5,6,7,8}, which is used for luma ALF, on CCALF adaptive precicion.
Cited:
Decisions
Study using the set of precisions {5,6,7,8}, which is used for luma ALF, on CCALF adaptive precicion.
Citation
Abstract
This contribution is a crosscheck report of JVET-AG0065 (Non-EE2: Adaptive coefficient precision for CCALF). The simulations were carried out based on the common test condition. The partial results are matched in BD-rate. Timing is not reliable in the crosscheck report.
Loading full meeting notes...
JVET-AG0289 Crosscheck of JVET-AG0065 (Non-EE2: Adaptive Coefficient Precision for CCALF) [N. Hu (Qualcomm)] [late]
References:
Decisions
JVET-AG0289 Crosscheck of JVET-AG0065 (Non-EE2: Adaptive Coefficient Precision for CCALF) [N. Hu (Qualcomm)] [late]
Citation
Abstract
This contribution reports the crosscheck results of JVET-AG0065[1]. In JVET-AG0065, the adaptive coefficient precision for CCALF is proposed. The encoder could select the best coefficient precision for each online-trained CCALF filter set and signal the selected coefficient precision in APS.
Loading full meeting notes...
JVET-AG0264 Crosscheck of JVET-AG0065 (Non-EE2: Adaptive coefficient precision for CCALF) [W. Yin (Bytedance)] [late]
References:
Decisions
JVET-AG0264 Crosscheck of JVET-AG0065 (Non-EE2: Adaptive coefficient precision for CCALF) [W. Yin (Bytedance)] [late]
Citation
Abstract
This document provides a draft call for video training materials intended for usage in the development of neural network-based video coding technologies.
Loading full meeting notes...
JVET-AG2036 Call for training materials for neural network-based video coding tool development, also issued as WG 5 N 266
The following 2 documents were produced as WG 5 documents only, without a corresponding JVET output document or direct repetition of their content in this meeting report:
WG 5 N 262 Request for subdivision: ISO/IEC TR 23888-3 Optimization of encoders and receiving systems for machine analysis of coded video content
WG 5 N 269 Liaison statement to ISO/IEC JTC 1/SC 29/WG 1 (JPEG) on JPEG AI and explorations on video coding
For the organization and planning of its future work, the JVET established 16 “ad hoc groups” (AHGs) to progress the work on particular subject areas. At this meeting, 2 Exploration Experiments (EE) were defined. The next eight JVET meetings were planned for 17 – 24 April 2024 under ITU-T SG16 auspices in Rennes, FR; during 12 – 19 July 2024 under ISO...
Cited:
Decisions
a) JVET documents
The meeting was closed at 2358 UTC on 2024-01-26.
Citation
Abstract
In adaptive loop filter of ECM-11.0, when filtering a luma sample with an on-line derived filter, 7 bits are used to represent the fractional part of a coefficient. Similarly, when filtering a chroma sample with a cross component adaptive loop filter (CCALF), 7 bits are used to represent the fractional part of a coefficient. Each CCALF coefficient can only be a value in the form of power of 2. In test EE2-4.3, the number of bits used to represent the fractional part of a luma coefficient is adaptive from 5 to 8, inclusively. A two-bit syntax element is signalled for each set of luma filter set to indicate the number of bits. In this contribution, the idea of EE test is applied to CCALF, the number of bits used to represent the fractional part of a CCALF coefficient is adaptive from 7 to 10, inclusively. In addition, the power of 2 constraint is removed. The proposed method was implemented on top of ECM-11.0, with common test condition, simulation results are reported as follows.
Loading full meeting notes...
JVET-AG0233 EE2-related: Adaptive precision for CCALF coefficients [N. Hu, M. Karczewicz, V. Seregin, H. Wang (Qualcomm)] [late]
In adaptive loop filter of ECM-11.0, when filtering a luma sample with an on-line derived filter, 7 bits are used to represent the fractional part of a coefficient. Similarly, when filtering a chroma sample with a cross component adaptive loop filter (CCALF), 7 bits are used to represent the fractional part of a coefficient. Each CCALF coefficient can only be a value in the form of power of 2. In test EE2-4.3, the number of bits used to represent the fractional part of a luma coefficient is adaptive from 5 to 8, inclusively. A two-bit syntax element is signalled for each set of luma filter set to indicate the number of bits. In this contribution, the idea of EE test is applied to CCALF, the number of bits used to represent the fractional part of a CCALF coeffic...
Cited:
Decisions
In later meeting discussion, further notes were added to JVET-AG0065. See notes also under JVET-AG0065.
Citation
Abstract
This document reports the work of the JVET ad hoc group on enhanced compression beyond VVC capability (AHG12) between the 32nd meeting, Hannover, DE, 13–20 October 2023, and the 33rd meeting, by teleconference, 17–26 January 2024.
Loading full meeting notes...
JVET-AG0012 JVET AHG report: Enhanced compression beyond VVC capability (AHG12) [M. Karczewicz, Y. Ye, L. Zhang (co-chairs), B. Bross, R. Chernyak, X. Li, K. Naser, Y. Yu (vice-chairs)]
Activities
The primary activity of the AHG was the “Exploration experiment on enhanced compression beyond VVC capability” (JVET-AF2024). The combined improvements of the ECM-11.0 over VTM-11.0ecm11.0 anchor for AI, RA and LB configurations are:
All-Intra Main 10
Y
U
V
EncT
DecT
Class A1
-10.81%
-21.80%
-29.33%
859.1%
400.6%
Class A2
-16.95%
-28.77%
-31.78%
860.2%
405.4%
Class B
-11.31%
-26.84%
-25.03%
784.4%
411.0%
Class C
-11.41%
-16.72%
-17.24%
810.7%
381.9%
Class E
-15.01%
-24.77%
-23.07%
750.1%
431.9%
Overall
-12.81%
-23.73%
-24.81%
808.6%
405.0%
Class D
-9.48%
-14.56%
-14.65%
795.1%
383.1%
Class F
-26.60%
-35.72%
-35.58%
513.6%
392.7%
Class TGM
-39.53%
-47.43%
-46.69%
457.3%
461.2%
Random Access Main...
Decisions
The AHG recommended to review all the related contributions.
Citation