Video Codec Scale Mapping for Cross-Platform Decoding Accuracy
Find Innovative SolutionsGenerate 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
Engineering 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
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.
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.
2Productivity
If scale parameter values are quantized during encoding, then data transmission efficiency is improved, but reconstruction accuracy deteriorates
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.
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.
3Ease of operation
If different devices perform encoding and decoding independently, then device autonomy is improved, but parameter consistency deteriorates
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.
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.
Data Source
Figure 1~2
Figure 3~5
Figure 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).