Dual Tracking Units for Power-Constrained Subject Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems for subject tracking, particularly those using neural networks, consume high power and require significant computational resources, leading to battery drainage in portable devices like digital cameras during live view display.
Innovation Solution
An image processing apparatus with dual tracking units, one using deep learning (DL) and the other without, where the operational load and usage are adjusted based on image brightness and motion, allowing the system to switch between them to conserve power.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If subject tracking using neural network is applied to moving image for live view display, then tracking accuracy is improved, but power consumption increases and battery drains
Solution Approach 1:
The system dynamically switches between deep learning-based tracking and conventional tracking methods based on scene characteristics. The control unit determines whether to enable DL tracking by evaluating motion magnitude and subject detection results, adjusting the tracking approach in real-time to balance accuracy and power consumption.
Solution Approach 2:
The system changes operational parameters by disabling the deep learning tracking unit under specific conditions (low motion, successful subject detection). This parameter change (enabling/disabling DL processing) directly controls power consumption while maintaining tracking functionality through alternative methods when appropriate.
2Reliability
If deep learning tracking unit is always enabled, then tracking performance is maintained, but power consumption increases
Solution Approach 1:
The tracking system is segmented into two distinct units: a deep learning-based tracking unit for high-accuracy tracking and a conventional tracking unit for basic tracking. The control unit selectively activates the DL unit only when necessary, dividing the operational workload to conserve energy while maintaining performance when needed.
Solution Approach 2:
The system employs periodic evaluation of scene characteristics (motion magnitude, subject detection) to determine whether DL tracking should be active. This periodic assessment allows the system to switch between tracking modes based on current conditions, avoiding continuous DL processing and reducing energy loss.
3Use of energy by moving object
If conventional tracking without deep learning is used, then power consumption is reduced, but tracking accuracy deteriorates
Solution Approach 1:
The control unit acts as an intermediary that selects between deep learning tracking and conventional tracking based on scene analysis. It evaluates motion magnitude and subject detection results, then directs the appropriate tracking method to the tracking unit, ensuring accurate tracking when DL is needed while saving power when conventional methods suffice.
4Measurement precision
If deep learning tracking is used in low-light or panning scenes, then tracking accuracy may improve, but power consumption increases unnecessarily
Solution Approach 1:
The control unit uses feedback from motion detection and subject detection results to determine whether to enable deep learning tracking. By continuously monitoring scene characteristics and adjusting DL tracking activation accordingly, the system avoids unnecessary power consumption in scenarios where DL tracking provides minimal benefit, such as low-light or panning scenes with stable subjects.
Data Source
AI summary
An image processing apparatus includes a first tracking unit configured to perform subject tracking using an image obtained by an image capturing unit, a second tracking unit configured to perform subject tracking using the image obtained by the image capturing unit, an operational load in the second tracking unit being smaller than an operational load in the first tracking unit, and a control unit configured to control switching between enabling both the first tracking unit and the second tracking unit and disabling one of the first tracking unit and the second tracking unit based on brightness of the image obtained by the image capturing unit.