3D Multi-View Object Tracking With Periodic Error Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object tracking technologies struggle with errors and inaccuracies, particularly in crowded or occluded environments such as ball sports, leading to erroneous tracking results due to players or structures obstructing the view.
Innovation Solution
An information processing apparatus that combines multi-viewpoint frame analysis with both efficient tracking and identification methods, using identifiers to track object positions accurately by integrating tracking unit and identification unit processes, and correcting errors through complementary updates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If object tracking uses past tracking results and current state information to estimate current position, then tracking efficiency is improved, but tracking accuracy deteriorates in occluded or crowded environments
Solution Approach 1:
The patent implements feedback by periodically identifying the object from captured images and comparing the identified position with the tracked position. When discrepancies exceed a threshold, the system corrects the tracking result using the identified position as feedback, thereby maintaining accuracy while preserving tracking efficiency.
Solution Approach 2:
The system performs preliminary identification of the object from captured images at specific intervals before final tracking decisions are made. This preliminary action provides reference information that helps correct potential tracking errors before they accumulate, balancing efficiency with accuracy.
2Use of energy by moving object
If tracking continuously uses past results to estimate current position, then computational load is reduced, but error accumulation increases leading to erroneous tracking
Solution Approach 1:
The patent employs periodic action by performing full object identification from captured images at predetermined intervals rather than continuously. This periodic correction identifies and corrects accumulated tracking errors while maintaining low computational load during intermediate periods, thus preserving both efficiency and reliability.
Solution Approach 2:
The system performs self-correction by automatically detecting tracking errors through periodic identification and adjusting its own tracking results. The identification unit serves the tracking process by providing corrective feedback when needed, enabling the system to maintain reliability without continuous external intervention or heavy computational processing.
3Measurement precision
If full identification processing is performed continuously to ensure accuracy, then tracking precision is improved, but computational complexity increases
Solution Approach 1:
The patent reduces computational complexity by performing full identification processing only at predetermined intervals rather than continuously. During periodic intervals, the system executes comprehensive identification to correct tracking drift, while between intervals it relies on lighter-weight tracking estimation, thus achieving acceptable precision with reduced computational burden.
Solution Approach 2:
The system applies partial action by performing full identification processing only when necessary (at predetermined intervals or when error thresholds are exceeded) rather than continuously. This selective application of full identification maintains tracking precision where needed while avoiding unnecessary computational complexity during stable tracking periods.
Data Source
Figure 1
Figure 2A~2B
Figure 3
AI summary
An information processing apparatus(100) according to the present disclosure obtains a plurality of captured images obtained by capturing images of an object from different directions(S301), estimates a three-dimensional shape of the object using the plurality of captured images(S302), performs tracking processing of the object to track the object by estimating an identifier and a position of the object using the three-dimensional shape(S303), performs identification processing of the object to identify the object using at least some of the plurality of captured images(S311) and estimate an identifier and a position of the object, detects an error in the result of the tracking processing based on the result of the tracking processing and the result of the identification processing, and complements tracking data indicating the result of the tracking processing using the result of the identification processing in a case where an error in the tracking processing is detected(S315).