Subject Tracking With Partial-Region Association Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for associating the entirety and part of a tracking-target subject based on positional relationship fail when similar objects cross each other, leading to erroneous associations, as seen in Japanese Patent Laid-Open No. 2021-152578, particularly when the heads of individuals overlap.
Innovation Solution
An information processing apparatus that detects a subject and its specific part, tracks the subject, and associates the specific part using a state of change in the quantity of detected partial regions, employing a threshold-based control to inhibit erroneous associations by adjusting the determination index and threshold in accordance with the change in local region detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If association is performed based on positional relationship between subject and partial region, then tracking accuracy is improved, but erroneous associations occur when similar objects cross each other
Solution Approach 1:
The system uses feedback by comparing the quantity of detected partial regions in the current frame with the previous frame. When the quantity changes (decreases or increases), the system adjusts its association strategy accordingly, preventing erroneous associations while maintaining tracking accuracy through dynamic adaptation based on detection state feedback
Solution Approach 2:
The association mechanism transitions from a static positional relationship-based approach to a dynamic approach that adapts based on the change state of partial region detection quantity. The system dynamically switches between different association modes (normal association, inhibition of association, or candidate selection) depending on whether the detection quantity increases, decreases, or remains stable
2Reliability
If the quantity of detected partial regions changes due to overlapping objects, then detection robustness is improved, but association accuracy deteriorates without proper control
Solution Approach 1:
The system performs preliminary action by detecting changes in the quantity of partial regions before performing association. By identifying whether the detection quantity has increased or decreased in advance, the system can proactively adjust its association strategy to prevent erroneous associations, rather than reacting after errors occur
Solution Approach 2:
The system changes the association parameter (determination index threshold) based on the detection state. When the quantity of detected partial regions changes, the system modifies the threshold for determining valid associations, allowing it to maintain association accuracy despite variations in detection robustness caused by overlapping objects
Data Source
AI summary
There is provided with an information processing apparatus. A first detecting unit detects a subject from each of a first image and a second image that chronologically follows the first image. A second detecting unit detects, from each of the first and second images, a partial region showing a specific part relating to the subject. An acquiring unit acquires a state of a change, between the first and second images, in a quantity in which the partial region is detected. A first controlling unit, in accordance with the state of the change, performs control of whether or not to associate a specific part shown in the partial region extracted in the second image with the subject.