Scenario-Specific Body-Part Tracking Retraining
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
General-purpose body-part trackers face challenges in effectively tracking human subjects in specific scenarios due to the need for costly, time-consuming, and expertise-intensive retraining processes, often requiring human-supplied ground truth annotations and statistically representative training sets.
Innovation Solution
A method to retrain a general-purpose body-part tracker using scenario-specific data without supervision, allowing for the identification of special-purpose parameters that can augment or replace general-purpose parameters, enabling improved tracking of human subjects in particular scenarios without the need for extensive technical expertise or resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If general-purpose body-part tracker is retrained using supervised machine learning with ground truth annotations, then tracking precision for specific scenarios is improved, but retraining cost and time consumption increase
Solution Approach 1:
The system performs self-training by automatically generating training data from unannotated scenario-specific videos. The general-purpose tracker processes these videos to generate predictions, which are then used to create training datasets without requiring manual ground truth annotations, enabling the tracker to adapt to specific scenarios autonomously
Solution Approach 2:
The system creates synthetic training datasets by copying and processing unannotated scenario videos through the general-purpose tracker. These generated predictions serve as training data, replacing the need for expensive manual annotation while preserving the essential tracking patterns for the specific scenario
2Measurement precision
If general-purpose body-part tracker is retrained using supervised machine learning with ground truth annotations, then tracking precision for specific scenarios is improved, but retraining complexity and expertise requirements increase
Solution Approach 1:
The system eliminates the need for expert intervention in data annotation and dataset creation by automatically generating training data through self-processing of unannotated videos. This self-service approach simplifies the retraining workflow from a complex expert-driven process to an automated procedure
Solution Approach 2:
The system uses copying of unannotated scenario videos through the general-purpose tracker to generate training data, replacing the complex manual annotation process with an automated copying and processing workflow that requires minimal expertise
3Reliability
If scenario-specific parameters are selectively used to augment or replace general-purpose parameters, then tracking performance in specific scenarios is improved, but parameter management complexity increases
Solution Approach 1:
The parameter set is segmented into general-purpose parameters and scenario-specific parameters. The system selectively applies scenario-specific parameters only when processing videos from their corresponding scenarios, while using general-purpose parameters for all scenarios. This segmentation reduces the active parameter management complexity while maintaining high tracking performance
4Loss of time
If unsupervised retraining is performed using scenario data without ground truth annotations, then retraining cost and time are reduced, but initial tracking accuracy may decrease
Solution Approach 1:
The system performs preliminary processing of unannotated scenario videos through the general-purpose tracker to generate predictions that serve as training data. This preliminary action creates a foundation of scenario-specific training examples that guide subsequent self-training, enabling the tracker to adapt to the specific scenario while maintaining reasonable initial accuracy
Solution Approach 2:
The system implements a feedback loop where the general-purpose tracker processes unannotated videos to generate predictions, these predictions are used to create training data, and the tracker is retrained using this data. This iterative feedback process gradually improves tracking accuracy for the specific scenario without requiring manual annotations
Data Source
AI summary
A human subject is tracked within a scene of an observed depth image supplied to a general-purpose body-part tracker. The general-purpose body-part tracker is retrained for a specific scenario. The general-purpose body-part tracker was previously trained using supervised machine learning to identify one or more general-purpose parameters to be used by the general-purpose body-part tracker to track a human subject. During a retraining phase, scenario data is received that represents a human training-subject performing an action specific to a particular scenario. One or more special-purpose parameters are identified from the processed scenario data. The special-purpose parameters are selectively used to augment or replace one or more general-purpose parameters if the general-purpose body-part tracker is used to track a human subject performing the action specific to the particular scenario.