Image Tracking Model Switching for Masking Continuity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing tracking technologies struggle with maintaining accuracy when a tracking destination is partially masked by another object, leading to potential misidentification of the tracking target due to similar appearances of masking materials.
Innovation Solution
An image processing apparatus that switches between tracking models based on detection of masking and its termination, using a first model for the masking material when masking occurs and a second model for the original tracking target when masking ends, ensuring accurate tracking by adapting to changes in visibility.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the tracking model lowers the threshold value to detect the masking material as a temporary tracking target, then tracking can be continued when the tracking destination is masked, but when another object with similar appearance appears, the tracking target may transition to the wrong object
Solution Approach 1:
The tracking model dynamically changes its detection threshold based on the masking state. When masking is detected, the threshold is lowered to capture the masking material as a temporary target. When no masking is detected, the threshold returns to normal levels. This dynamic adjustment resolves the contradiction by adapting the detection sensitivity to the current scene conditions.
Solution Approach 2:
The invention changes the detection threshold parameter of the tracking model based on whether masking is detected. By switching between different threshold values (high for normal tracking, low for masking detection), the system maintains both tracking continuity during masking and tracking accuracy during normal conditions.
2Measurement precision
If a single tracking model is used for all conditions, then the device complexity is low, but the tracking accuracy deteriorates when masking occurs
Solution Approach 1:
A single tracking model performs multiple functions by switching between different detection thresholds. The same model structure handles both normal tracking and masking detection scenarios, eliminating the need for separate models while maintaining high tracking accuracy in both conditions.
Solution Approach 2:
Instead of using multiple different models, the invention changes the parameter (detection threshold) of a single tracking model to adapt to different conditions. This approach maintains model simplicity while achieving high tracking accuracy for both normal objects and masking materials.
Data Source
AI summary
An image processing apparatus comprises a tracking unit configured to perform a tracking process, using a tracking model, in which a tracking target in a captured image is tracked, and a switching unit configured to switch the tracking model to a first model that tracks a second object as the tracking target when masking of the first object by the second object is detected while the tracking unit tracks the first object as the tracking target, and to switch the tracking model to a second model that tracks the first object as the tracking target when termination of the masking is detected.