Object Tracking Device Candidate Likelihood Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object tracking techniques fail to distinguish between objects when they are near each other, leading to label mixing and frequent ID changes, resulting in fragmented tracking results.
Innovation Solution
An object tracking device and method that stores candidates with label information, position, and likelihood, where the likelihood of each candidate is controlled based on the label information of surrounding candidates, allowing for proper tracking even when objects are close together.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If candidates are generated and moved to detection results without considering label relationships, then the tracking process can handle unknown and changing numbers of objects, but labels mix together when objects are near each other causing indistinguishability
Solution Approach 1:
The patent applies local quality by adjusting the likelihood of each candidate based on the label information of neighboring candidates. Instead of treating all candidates uniformly, the system modifies likelihoods locally according to the spatial distribution and label characteristics of nearby candidates, thereby maintaining label distinction accuracy even when objects are close together while preserving the ability to track unknown numbers of objects
2Device complexity
If candidates are generated irrespective of label values, then the tracking system remains simple, but ID changes frequently causing fragmented tracking results
Solution Approach 1:
The patent implements feedback by using the label information of neighboring candidates to adjust the likelihood of each candidate. This feedback mechanism allows the system to automatically adjust tracking parameters based on the current state of candidates, preventing frequent ID changes and maintaining consistent tracking results without significantly increasing process complexity
Data Source
AI summary
Tracking candidate storage means 81 stores a plurality of candidates for one or more objects to be tracked. Control means 82 controls a likelihood of each of the plurality of candidates. The tracking candidate storage means 81 stores, for each of the plurality of candidates, label information identifying which object the candidate corresponds to, position information indicating a position of the object, and a likelihood. The control means 82 controls the likelihood of a candidate, according to the label information of one or more other candidates around the candidate determined based on the position information of the candidate.


