Dynamic Gating of Transformed Domain Image Components
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing methods for artificial intelligence tasks consume significant communication and processing resources due to the conversion between native and transformed domain formats, leading to increased latency and power consumption.
Innovation Solution
A method that dynamically gates encoded image components based on their importance and available computational resources, allowing only essential components to be processed, thereby reducing the need for frequent format conversions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If image data is converted to transformed domain format for transmission and storage, then data compression and bandwidth efficiency are improved, but additional processing steps are required to convert back to native format for AI tasks
Solution Approach 1:
The patent extracts and removes the inverse transformation step from the conventional processing pipeline. By enabling AI tasks to operate directly on transformed domain data (DCT coefficients, quantized values) without converting back to native RGB/YCbCr format, it eliminates redundant processing while maintaining task effectiveness through selective component gating.
Solution Approach 2:
The patent segments the image data processing by selectively gating individual frequency components or blocks of transformed domain data. Instead of processing all components uniformly or converting the entire image, it identifies and processes only the most important components for specific AI tasks, reducing overall processing complexity.
2Reliability
If all transformed domain components are processed for AI tasks, then task accuracy is maintained, but computational workload and processing time increase
Solution Approach 1:
The patent applies local quality by treating different frequency components or image blocks differently based on their importance to the specific AI task. Important components are processed with high fidelity while less important ones are gated out or processed at lower precision, optimizing the balance between accuracy and efficiency for each local region or frequency band.
Solution Approach 2:
The patent implements partial action by processing only a subset of transformed domain components rather than all components. Through gating mechanisms, it selectively activates processing for components that contribute most to task accuracy, performing 'enough' processing to achieve required performance without the excessive computation of complete processing.
3Adaptability or versatility
If frequent format conversions are performed between native and transformed domains, then data compatibility is ensured, but processing latency and power consumption increase
Solution Approach 1:
The patent removes the inverse transformation step from the processing pipeline entirely. By designing AI tasks that natively operate on transformed domain representations, it extracts and eliminates the redundant conversion step that causes latency and power consumption, while maintaining compatibility through standardized transformed domain interfaces.
4Speed
If computational resources are increased to process all image components, then processing speed is improved, but energy consumption and system cost increase
Solution Approach 1:
The patent applies partial action by processing only the essential subset of image components required for each AI task. Through dynamic gating that identifies and processes only the most informative components, it achieves required processing speed without the excessive energy consumption that would result from processing all components with full computational resources.
Data Source
AI summary
A system for processing encoded image components for artificial intelligence tasks. The system can include one or more compute units, one or more controllers and memory. The one or more controllers can include one or more micro-op schedulers and one or more channel switches. The one or more compute units can be configured to process components of the transformed domain image data according to one or more micro-operations for an artificial intelligence task. The one or more channel switches can be configured to selectively control the transfer of the components of transformed domain image data to the one or more compute units based on one or more gating flags. The one or more channel switches can also be configured to selectively control generation of the one or more micro-operations by the one or more micro-op schedulers based on the one or more gating flags.