Usage-Aware Video Encoding for Human and AI Perception
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image compression technologies are not optimized for artificial intelligence services, lacking efficiency and suitability for machine tasks, and there is a need for improved encoding/decoding methods that consider image usage and machine perception.
Innovation Solution
An encoding/decoding method and apparatus that optimize image data based on its usage, determining data optimization information for human or machine tasks, and encoding this information into a bitstream for efficient transmission and decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If existing image compression technologies are used for high-resolution and high-quality image processing for human vision, then image quality is improved, but they are not suitable for artificial intelligence services
Solution Approach 1:
The patent introduces dynamic optimization by allowing the image compression system to adapt its encoding parameters based on the intended usage. The decoder receives usage information (human perception or machine task) and dynamically adjusts decoding parameters to optimize for the specific application, making the system versatile for both high-quality human viewing and AI processing tasks
Solution Approach 2:
The patent changes key parameters of the image compression system based on usage type. For human perception, it optimizes for visual quality metrics, while for machine tasks, it optimizes for task performance metrics. This parameter adaptation allows the same compression framework to achieve both high image quality for humans and high effectiveness for AI services
2Manufacturing precision
If image compression is optimized for human vision, then human perception quality is improved, but encoding/decoding efficiency for machine tasks deteriorates
Solution Approach 1:
The patent segments the image processing pipeline into distinct optimization paths: one for human perception and another for machine tasks. The system receives usage information that determines which path to follow, allowing specialized optimization for each segment without compromising the other. This segmentation enables efficient machine task processing when that usage is detected
3Adaptability or versatility
If general-purpose image compression is used, then broad applicability is improved, but specialized optimization for specific usage deteriorates
Solution Approach 1:
The patent creates a universal image compression framework that can serve multiple functions by incorporating usage information handling. The same compression core processes images for both human perception and machine tasks, with the optimization parameters adapting based on the detected usage type. This multi-functionality maintains broad applicability while achieving specialized optimization performance
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present disclosure relates to an encoding/decoding method and apparatus for optimizing quantization according to video usage, a method for transmitting a bitstream, and a bitstream storage medium. A decoding method according to an embodiment of the present disclosure may comprise the steps of: acquiring data optimization information related to a video data optimization method according to the purpose of video data; and reconstructing the video data on the basis of the data optimization information, wherein the data optimization information includes data optimization application information indicating whether optimization is performed for each usage of the video data.