Neural Thumbnail Decoding From Feature Maps to Cut Display Lag
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for obtaining video thumbnails are complex and time-consuming, affecting the speed at which digital video applications display original pictures.
Innovation Solution
A method involving decoding a bitstream to obtain a lower-resolution feature map and reconstructing a target-resolution picture using neural networks, bypassing the need to first reconstruct the original picture, thereby improving efficiency and speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If the original picture is first decoded and reconstructed to obtain the original picture, then the thumbnail can be obtained by reducing the resolution, but the process becomes complex and consumes a long period of time
Solution Approach 1:
The patent extracts the essential feature information from the original picture through feature extraction, obtaining a feature map that captures the key visual characteristics. This feature map is then used to directly generate the thumbnail through neural network reconstruction, skipping the unnecessary step of reconstructing the complete original picture. This extraction approach reduces processing time while maintaining thumbnail quality.
Solution Approach 2:
The patent segments the picture processing into two independent stages: feature extraction stage (original picture → feature map) and thumbnail generation stage (feature map → thumbnail). By separating these functions and using the feature map as an intermediate representation, the system avoids the time-consuming full picture reconstruction while preserving the essential visual information needed for thumbnail generation.
2Manufacturing precision
If the original picture is first decoded and reconstructed, then the thumbnail can be generated, but the complexity of the process increases
Solution Approach 1:
The patent introduces a feature map as an intermediary representation between the original picture and the thumbnail. Instead of directly processing the original picture through complex reconstruction and rescaling operations, the system uses the feature map as a compact intermediate that captures essential visual information. This intermediary approach simplifies the overall processing pipeline while maintaining output quality.
Solution Approach 2:
The patent changes the representation parameters from pixel-based original pictures to feature-map-based abstract representations. By transforming the data from high-resolution pixel values to compressed feature vectors, the system reduces the dimensional complexity and enables more efficient processing. The neural network then operates on these simplified feature parameters to generate the thumbnail, significantly reducing computational complexity.
3Productivity
If conventional decoding methods are used to generate thumbnails, then the process can be implemented, but frame freezing occurs due to slow processing speed
Solution Approach 1:
The patent replaces conventional mechanical image processing (decoding → reconstructing → rescaling) with a neural network-based system that directly transforms feature maps to thumbnails. This substitution of the processing mechanism enables parallel computation and faster inference, dramatically improving thumbnail generation speed and preventing frame freezing in video playback.
Solution Approach 2:
The patent performs feature extraction as a preliminary action on the original picture, creating a feature map that can be quickly processed to generate thumbnails. By preparing the feature representation in advance, the system enables rapid thumbnail generation without waiting for complete picture decoding and reconstruction, thus maintaining smooth frame display continuity.
Data Source
AI summary
A decoding method includes decoding a bitstream to obtain a first feature map, where a resolution of the first feature map is lower than a resolution of an original picture; and reconstructing a second feature map based on a first neural network, to obtain a reconstructed picture, where a resolution of the second feature map and a resolution of the reconstructed picture each are a target resolution, the target resolution is lower than the resolution of the original picture, and the second feature map is the first feature map, or the second feature map is a feature map obtained by processing the first feature map based on a second neural network.