Video Codec Scale Mapping for Cross-Platform Decoding Accuracy

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In cross-platform video transmission, the accuracy of video frames obtained by decoding and reconstruction is relatively low due to inconsistencies in encoding and decoding processes performed by different computer devices.

Innovation Solution

A video encoding and decoding model processing method that includes obtaining a training video frame, extracting a feature map, determining scale parameter values, mapping them to a preset range, and updating model parameters based on quantization constraint loss to improve accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If video encoding and decoding are performed by different computer devices in cross-platform transmission, then video transmission compatibility is improved, but decoding accuracy deteriorates

Engineering Contradiction:
Improvecross-platform compatibilityVSAvoiddecoding accuracy
Core Design Contradiction:
Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent changes the parameter distribution of scale values by introducing a constraint loss function that penalizes scale values close to rounding boundaries. This parameter transformation ensures that scale values are pushed away from boundaries, reducing quantization inconsistencies across different platforms while maintaining cross-platform compatibility.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent applies preliminary action by pre-training the encoding and decoding models together with a constraint loss function before actual video transmission. This pre-training process establishes consistent parameter distributions across different platforms, preventing decoding accuracy deterioration from the outset.

Inventive Principle:
Principle #10Preliminary action

2Productivity

If scale parameter values are quantized during encoding, then data transmission efficiency is improved, but reconstruction accuracy deteriorates

Engineering Contradiction:
Improvedata transmission efficiencyVSAvoidreconstruction accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent applies preliminary anti-action by introducing a constraint loss function that anticipates and counteracts the negative effects of quantization. The loss function penalizes scale values that would round to the same integer, thereby pre-preventing potential accuracy losses before they occur during the quantization process.

Inventive Principle:
Principle #9Preliminary anti-action

Solution Approach 2:

The patent transforms the distribution of scale parameters through the constraint loss function, shifting parameters away from rounding boundaries. This parameter change ensures that quantized values maintain better correspondence between encoding and decoding across different platforms, improving reconstruction accuracy while preserving transmission efficiency.

Inventive Principle:
Principle #35Parameter changes

3Ease of operation

If different devices perform encoding and decoding independently, then device autonomy is improved, but parameter consistency deteriorates

Engineering Contradiction:
Improvedevice autonomyVSAvoidparameter consistency
Core Design Contradiction:
Ease of operationVSStability of the object's composition

Solution Approach 1:

The patent uses preliminary action by jointly training encoding and decoding models with a constraint loss function before deployment. This pre-training establishes consistent parameter distributions and rounding behaviors across different devices, ensuring parameter consistency is maintained even when devices operate independently afterward.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent implements feedback through the constraint loss function that monitors and penalizes parameter inconsistencies during model training. The loss function provides continuous feedback to adjust scale parameter distributions, ensuring that encoding and decoding parameters remain consistent across different devices while maintaining their operational autonomy.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP4723629A1Video codec model processing method, video encoding method, video decoding method and related device
Publication Date: 2026.04.08 TENCENT TECHNOLOGY (SHENZHEN) CO LTD
  • EP4723629A1 patent drawingFigure 1~2
  • EP4723629A1 patent drawingFigure 3~5
  • EP4723629A1 patent drawingFigure 6~7

AI summary

A video codec model processing method, executed by a computing device and comprising : acquiring a training video frame, extracting a feature map of the training video frame by means of a video codec model, and determining scale parameter values respectively corresponding to feature elements in the feature map (202); on the basis of a preset mapping relationship, respectively mapping the scale parameter values corresponding to the feature elements to obtain scale parameter mapping values respectively corresponding to the feature elements, the scale parameter mapping values being within a preset mapping value range (204); acquiring a constraint reference mapping value corresponding to each scale parameter mapping value, wherein the distance between the constraint reference mapping value and a rounding boundary value of the scale parameter mapping value corresponding to the constraint reference mapping value is greater than the distance between the scale parameter mapping value corresponding to the constraint reference mapping value and the rounding boundary value (206); on the basis of the difference between the scale parameter mapping value corresponding to each feature element and the constraint reference mapping value corresponding to each scale parameter mapping value, determining a quantization constraint loss of the training video frame (208); and updating at least some of model parameters of the video codec model on the basis of the quantization constraint loss, so as to obtain a trained video codec model (210).