Motion Capture Group Tracking with Occlusion Data Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current motion capture systems face challenges in effectively tracking and interpreting the movements of multiple individuals within a field of view, particularly when individuals are occluded or partially out of view, leading to discontinuous and unrealistic interactions with applications.
Innovation Solution
A processor-implemented method that tracks a group of people by detecting their bodies, determining their movements as a unitary entity, and generating missing data for occluded individuals, while using audio inputs to provide a seamless and natural interaction experience with applications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If motion capture systems track multiple individuals separately, then individual movement data can be captured, but the system fails to capture group dynamics and produces discontinuous output when individuals are occluded
Solution Approach 1:
The patent merges individual tracking data into a unified group representation by detecting the collective bounding box of all individuals and using it to generate missing data for occluded persons. This combining approach maintains tracking continuity by relying on group-level information rather than requiring complete individual detection at all times.
Solution Approach 2:
The patent introduces an intermediate group representation (collective bounding box and group center) that mediates between individual detection failures and continuous tracking output. This intermediary structure allows the system to infer individual positions from group-level data when individuals are occluded, ensuring reliable continuous tracking.
2Reliability
If the system generates missing data for occluded individuals, then tracking continuity is maintained, but the complexity of data processing increases
Solution Approach 1:
The system performs preliminary actions by continuously maintaining the collective bounding box and group center information during tracking. This pre-computed group-level data is readily available when individuals become occluded, enabling quick generation of missing data without complex real-time calculations.
Solution Approach 2:
The patent uses copying by replicating the group-level bounding box and center information to generate missing individual data. Instead of complex inference, the system copies the collective representation and uses it to reconstruct individual trajectories, simplifying the data generation process while maintaining continuity.
3Adaptability or versatility
If the system uses group movement as unitary input, then natural group interaction is captured, but individual control precision is reduced
Solution Approach 1:
The patent implements multi-functionality by enabling the motion capture system to operate in multiple modes: individual tracking when persons are visible, and group-level tracking when persons are occluded. The system universally handles both individual and group interactions, adapting its precision level based on visibility conditions.
Data Source
AI summary
In a motion capture system, a unitary input is provided to an application based on detected movement and/or location of a group of people. Audio information from the group can also be used as an input. The application can provide real-time feedback to the person or group via a display and audio output. The group can control the movement of an avatar in a virtual space based on the movement of each person in the group, such as in a steering or balancing game. To avoid a discontinuous or confusing output by the application, missing data can be generated for a person who is occluded or partially out of the field of view. A wait time can be set for activating a new person and deactivating a currently-active person. The wait time can be adaptive based on a first detected position or a last detected position of the person.