Computer Vision Processing Power Reduction via Frame Change Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computer vision applications in wireless electronic devices consume significant power due to the need for application-specific processing on each frame of a video stream, even when frames are similar, leading to resource-intensive computations and battery drain.
Innovation Solution
A system that generates image statistics for each frame and uses a change detection circuit to determine if differences between frames satisfy a threshold, allowing for the omission of processing on similar frames and activating a high-power processing path only when necessary, thereby conserving power by using a hierarchy of processing resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If application-specific processing is performed on each frame of video stream, then computer vision application accuracy is improved, but power consumption increases
Solution Approach 1:
The system performs preliminary change detection between frames using a change detection circuit before committing to full application-specific processing. This preliminary action identifies frames that are substantially similar to previous frames, allowing the system to skip processing for those frames and thereby reduce power consumption while maintaining accuracy for frames that actually contain changes
Solution Approach 2:
Instead of performing complete application-specific processing on every frame, the system applies partial processing only to frames that exhibit significant changes from previous frames. This selective approach processes only the necessary subset of frames, reducing overall power consumption while maintaining sufficient accuracy for detecting actual events or objects in the video stream
2Reliability
If computer vision processing is performed on every frame, then detection reliability is improved, but resource utilization becomes inefficient
Solution Approach 1:
The system performs preliminary change detection using a change detection circuit and image statistics comparison before full processing. This preliminary filtering identifies frames worth processing, improving resource utilization efficiency by avoiding unnecessary computation on redundant frames while maintaining detection reliability for frames that contain actual changes
Solution Approach 2:
The system dynamically adjusts processing intensity based on frame-to-frame changes. When changes exceed a threshold, full application-specific processing is applied; when changes are minimal, processing is skipped or reduced. This dynamic adaptation optimizes resource utilization efficiency while preserving detection reliability for meaningful events
3Loss of information
If all images are processed for visual descriptors, then feature completeness is improved, but computational cost increases
Solution Approach 1:
The system extracts visual descriptors only for frames that exhibit significant changes from previous frames, rather than processing all frames. This partial extraction approach reduces computational cost and power consumption while maintaining sufficient feature completeness for frames that contain actual changes or events of interest
Data Source
AI summary
Methods, systems, and techniques to enhance computer vision application processing are disclosed. In particular, the methods, systems, and techniques may reduce power consumption for computer vision applications and improve processing efficiency for computer vision applications.