Vision-Based Target Tracking With Face and Body Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vision-based target tracking algorithms, particularly those using correlation filtering, struggle with accurately tracking targets that exhibit abrupt changes in motion or posture, leading to poor tracking performance.
Innovation Solution
A vision-based target tracking method that combines time-series tracking algorithms with face and body recognition to improve tracking accuracy. This method involves reading current image frames to detect face and body positions, using time-series tracking to predict target positions, and employing face and body recognition to verify target identities and correct tracking errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If correlation filtering algorithm is used for target tracking, then real-time template matching is achieved, but tracking accuracy deteriorates under abrupt motion and posture changes
Solution Approach 1:
The patent segments the tracking process into multiple independent modules: face detection module, body detection module, time-series tracking module, face matching module, body matching module, and update module. Each module handles specific aspects of tracking, allowing the system to maintain real-time performance while improving accuracy through specialized processing of different body parts and temporal information.
Solution Approach 2:
The patent performs preliminary actions by detecting faces and bodies in advance, storing their position information, and using time-series tracking to predict future positions before actual tracking occurs. This preparation allows the system to handle abrupt changes more effectively by having predictive models ready.
Solution Approach 3:
The patent implements feedback mechanisms where face matching and body matching results are fed back to the update module, which then corrects tracking errors and updates the target position. This continuous feedback loop ensures that tracking accuracy is maintained even under abrupt motion and posture changes.
2Reliability
If historical templates are optimized for robust tracking, then template reliability is improved, but response time to abrupt changes deteriorates
Solution Approach 1:
The patent makes the tracking system dynamic by using time-series tracking algorithms that adapt to changing motion patterns in real-time. Instead of relying on static historical templates, the system continuously updates its understanding of target motion based on current video frames, enabling fast response to abrupt changes while maintaining robustness through learned temporal patterns.
Solution Approach 2:
The system performs preliminary action by using time-series tracking to predict target positions before actual changes occur. The predictive model, trained on historical data, prepares advance information about expected motion patterns, allowing the system to respond quickly to abrupt changes without waiting for template optimization.
3Measurement precision
If face and body recognition is added to verify target identity, then tracking accuracy is improved, but system complexity increases
Solution Approach 1:
The patent segments the complex verification task into separate face matching and body matching modules, each handling specific aspects of identity verification. This segmentation allows the system to achieve high accuracy through specialized processing while keeping each individual module relatively simple and manageable.
Solution Approach 2:
The patent merges face detection, body detection, time-series tracking, face matching, and body matching into a unified tracking system coordinated by an update module. This integration allows the system to achieve high accuracy through multiple verification mechanisms while managing complexity through coherent architecture and shared data structures.
Data Source
AI summary
A vision-based target tracking method, a system, an equipment, and a storage medium. The method includes: acquiring face positions and body positions in a current image frame; acquiring a current position of a time-series tracking target by a time-series tracking algorithm; determining, when a face matching the face of a to-be-tracked target does not exist outside a region corresponding to the current position of the time-series tracking target, whether the time-series tracking target is in an intersecting state; determining whether a body matching the body of the to-be-tracked target in other persons intersecting the times-series tracking target when in the intersecting state, or using the current position of the time-series tracking target as a current target position if not in the intersecting state; and taking the current target position as the position of the to-be-tracked target in the previous frame, and returning to the above steps.