Drift Determining Model for Target Tracking Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing target tracking methods suffer from tracker drift due to factors like blocking, motion blurring, and appearance changes, leading to inaccurate tracking and high computational costs, especially in real-time applications.
Innovation Solution
A drift determining model based on largest sample response values is used to quickly detect and correct tracker drift, improving tracking accuracy by modeling the largest sample response values and employing a coarse-fine process for drift correction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a detection algorithm is added to the tracker to resolve tracker drift, then tracking accuracy is improved, but computational complexity increases and real-time performance deteriorates
Solution Approach 1:
The patent extracts only the essential drift detection functionality from a full detection algorithm, creating a simplified drift determining model that checks whether the target remains within the search window based on largest sample response values. This extracted lightweight mechanism provides drift detection capability without the computational burden of complete detection algorithms.
Solution Approach 2:
The tracking system is segmented into distinct functional components: the tracker for fast positioning and the drift determining model for accuracy verification. This segmentation allows each component to be optimized independently - the tracker for speed and the drift model for accuracy - while working together to resolve the contradiction between reliability and complexity.
2Reliability
If a detection algorithm is added to the tracker to resolve tracker drift, then tracking accuracy is improved, but processing speed decreases
Solution Approach 1:
The patent extracts only the essential drift detection functionality from a full detection algorithm, creating a simplified drift determining model that checks whether the target remains within the search window based on largest sample response values. This extracted lightweight mechanism provides drift detection capability without the computational burden of complete detection algorithms.
Solution Approach 2:
Instead of applying full detection algorithms to every frame, the patent applies partial action by using the drift determining model only when drift is suspected (when largest sample response values indicate potential drift). This selective application maintains high processing speed while providing accuracy verification when needed.
3Adaptability or versatility
If multiple tracked targets are processed, then coverage is improved, but computational complexity increases excessively
Solution Approach 1:
The tracking system is segmented into distinct functional components: the tracker for fast positioning and the drift determining model for accuracy verification. This segmentation allows each component to be optimized independently - the tracker for speed and the drift model for accuracy - while working together to resolve the contradiction between reliability and complexity.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
This application provides a target tracking method, including: receiving a current frame of picture, where the current frame of picture includes a target object; determining, based on a drift determining model, whether a tracker drifts for tracking of the target object in the current frame of picture, where the drift determining model is obtained through modeling based on largest values of responses of samples; the drift determining model is a probability density model; the sample is a training sample used to train the drift determining model, where the training sample is collected from a training picture, and the training picture includes the target object; and the response value of the sample is a probability value of likelihood that the training sample is the target object in the training picture; and outputting a tracking drift result, where the tracking drift result includes: drift is generated for the tracking of the target object, or no drift is generated for the tracking of the target object.