Video Object Detection Energy Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object detection methods in video images are computationally intensive, requiring specialized GPUs for real-time operation with high accuracy, while object tracking is less efficient when dealing with multiple objects.
Innovation Solution
A method that determines the degree of correlation between images and decides whether to perform object detection or tracking based on energy consumption measurements, switching between the two processes to optimize energy use.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If object detection is performed on every image, then detection accuracy is maintained, but energy consumption increases significantly
Solution Approach 1:
The system dynamically switches between object detection and object tracking modes based on the degree of correlation between consecutive images. When correlation is high (images are similar), the system uses low-complexity tracking. When correlation is low (images differ significantly), the system uses high-accuracy detection. This dynamic adaptation resolves the contradiction by selecting the appropriate processing mode according to actual image conditions.
Solution Approach 2:
The system changes the processing parameter (detection vs. tracking) based on the correlation degree parameter. By measuring the correlation between consecutive images and using this to determine the processing mode, the system optimizes energy consumption while maintaining detection accuracy. The correlation degree serves as a threshold-based parameter that triggers mode switching.
2Use of energy by moving object
If object tracking is used for all images, then energy consumption is reduced, but detection accuracy deteriorates when images differ significantly
Solution Approach 1:
The system uses the correlation degree as feedback information to determine whether to perform detection or tracking. The correlation measurement provides feedback about image similarity, which then guides the selection of the appropriate processing mode. This feedback mechanism ensures that tracking is only used when image conditions justify it, preventing accuracy deterioration.
Solution Approach 2:
The system dynamically adjusts its behavior based on real-time correlation analysis. Instead of using a fixed tracking approach, it adapts its processing mode (detection or tracking) according to the actual similarity between consecutive images, ensuring both energy efficiency and accuracy are maintained appropriately.
3Productivity
If correlation analysis is performed for every image pair, then processing decisions are optimized, but computational complexity increases
Solution Approach 1:
The system performs correlation analysis as a preliminary step before the main processing (detection or tracking). This preliminary correlation measurement is a lightweight operation that guides subsequent processing decisions. By doing this analysis upfront, the system avoids unnecessary heavy processing while maintaining optimized decision-making.
Solution Approach 2:
The system applies correlation analysis selectively rather than exhaustively. Instead of analyzing all possible image pairs or using complex correlation methods always, it uses a simplified correlation measurement approach that provides sufficient information for processing decisions without excessive computational overhead.
Data Source
AI summary
A method of detecting objects in a video includes, for each image in the video, determining a degree of correlation between the image and at least one preceding image in a sequence of images; and obtaining a first measure of energy consumption required for performing an object detection process on the image, obtaining a second measure of energy consumption required for performing an object tracking process for tracking at least one object from a previous image, and comparing the first and second measures of energy. The object detection process is performed on the image if either said degree of correlation is below a threshold level or said first measure of energy is lower than the second measure of energy, and otherwise the object tracking process is performed on the image.