Neural Network Post-Filter Energy Adaptation for Video QoE

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding systems fail to efficiently adapt to energy consumption changes, leading to inefficient energy use and quality of experience.

Innovation Solution

Implementing a neural network post-filter for energy adaptation using a neural-network post-filter characteristic (NNPFC) SEI message to adjust energy consumption based on quality of experience (QoE) and user profiles.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Use of energy by moving object

If video coding systems use traditional compression methods, then storage and transmission bandwidth are reduced, but energy consumption cannot be efficiently adapted to different quality requirements

Engineering Contradiction:
Improveenergy consumptionVSAvoidadaptability to quality requirements
Core Design Contradiction:
Use of energy by moving objectVSAdaptability or versatility

Solution Approach 1:

The system dynamically adapts video quality by adjusting the neural network post-filter processing level based on real-time energy availability and quality requirements. The decoder can switch between different processing modes (full neural network processing, reduced processing, or no processing) to balance energy consumption against quality of experience, making the system flexible rather than static

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent changes key parameters including the neural network processing intensity, filter application level, and quality metric thresholds to optimize the balance between energy consumption and video quality. By adjusting these parameters based on energy adaptation indications, the system can operate efficiently across different energy scenarios while maintaining acceptable quality

Inventive Principle:
Principle #35Parameter changes

2Reliability

If neural network post-filter is applied to improve video quality, then quality of experience increases, but energy consumption increases

Engineering Contradiction:
Improvequality of experienceVSAvoidenergy consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

Instead of always applying full neural network post-filter processing, the system applies partial processing only when energy is available and quality improvement is needed. The energy adaptation mechanism allows the decoder to apply the neural network filter at reduced intensity or skip it entirely during energy-constrained periods, achieving acceptable quality with less energy expenditure

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The system uses feedback from energy adaptation indications and quality metric evaluations to dynamically control neural network post-filter application. The decoder monitors energy availability and quality requirements, then adjusts the level of neural network processing accordingly, creating a closed-loop system that balances quality and energy consumption

Inventive Principle:
Principle #23Feedback

3Reliability

If video quality is maintained at high levels, then quality of experience is improved, but energy waste increases

Engineering Contradiction:
Improvevideo qualityVSAvoidenergy waste
Core Design Contradiction:
ReliabilityVSLoss of energy

Solution Approach 1:

The system applies quality enhancement selectively rather than uniformly across all video content. The neural network post-filter is applied only to specific regions or time periods where quality improvement provides the most value, while reducing or eliminating processing in areas where high quality is less critical, thereby reducing overall energy waste while maintaining acceptable video quality

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP4676040A1Neural network post filter for energy adaptation
Publication Date: 2026.01.07 INTERDIGITAL CE PATENT HOLDINGS SAS
  • EP4676040A1 patent drawingFigure 1A
  • EP4676040A1 patent drawingFigure 1B
  • EP4676040A1 patent drawingFigure 1C

AI summary

Systems, methods, and instrumentalities are disclosed associated with neural network post filter for energy adaptation. Messages associated with energy adaptation operations may be signaled and/or used. The messages may be associated with neural network post-filter characteristics (NNPFC) and/or supplemental enhancement information (SEI). The message may be an NNPFC SEI message. Adaptation in terms of energy consumption of content (e.g., decoded content) may be performed, for example, based on a received NNPFC SEI message. An NNPFC SEI message may enable reduction of energy consumption of a content while used. An NNPFC SEI message may enable increase of energy consumption of a content, for example, if the increase in energy is associated with an increased quality of experience (QoE) associated with the content.