Subject Tracking with Adaptive Discriminator Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing tracking techniques using online learning suffer from reduced accuracy when there is insufficient training data or sessions, leading to instability and erroneous tracking.
Innovation Solution
An information processing apparatus with multiple discriminators, including one performing online learning, selectively uses a second discriminator when online learning is incomplete, integrating likelihood maps based on completeness evaluation to suppress errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If online learning is performed to improve tracking accuracy, then tracking accuracy is improved when sufficient training data is available, but tracking accuracy deteriorates when training data or sessions are insufficient
Solution Approach 1:
The system dynamically switches between offline learning discriminator and online learning discriminator based on the completeness of online learning. The tracking unit adapts its behavior by selecting which discriminator to use at each step, making the system flexible and responsive to the current state of learning completeness rather than being static
Solution Approach 2:
The system changes the parameter of discriminator selection based on the completeness of online learning. When completeness is high, the online learning discriminator is used; when completeness is low, the offline learning discriminator is used. This parameter change allows the system to optimize tracking accuracy based on data availability
2Reliability
If only offline learning discriminator is used, then tracking stability is maintained, but tracking accuracy cannot be improved through adaptive learning
Solution Approach 1:
The system performs preliminary offline learning to establish a stable baseline tracking performance before online learning begins. This preliminary action ensures that the tracking unit has a reliable foundation to work from, and the offline learning discriminator remains available as a fallback to maintain stability
Solution Approach 2:
The system uses feedback from the completeness evaluation to determine which discriminator to use. The tracking accuracy is continuously monitored and the discriminator selection is adjusted based on this feedback, allowing the system to improve accuracy while maintaining stability through adaptive decision-making
3Measurement precision
If multiple discriminators are used to improve tracking accuracy, then tracking accuracy is enhanced through selective use, but device complexity increases
Solution Approach 1:
The system extracts and uses only the necessary discriminator at each step based on the completeness of online learning. Rather than always using both discriminators or requiring a complex ensemble method, the system selectively extracts and applies the most appropriate discriminator for the current situation, reducing unnecessary complexity
Solution Approach 2:
Both the offline learning discriminator and online learning discriminator are designed to perform the same tracking function, but with different learning approaches. This multi-functionality allows the system to switch between them based on needs without requiring entirely separate systems, reducing overall complexity while maintaining accuracy
Data Source
AI summary
There is provided with an information processing apparatus. A tracking unit tracks a subject in an input image using one or both of a first discriminator that tracks the subject and a second discriminator that tracks the subject and is different from the first discriminator. An obtaining unit obtains training data used for training to track with the first discriminator. A learning unit performs online learning of causing the first discriminator to learn while tracking the subject using the training data. An evaluating unit evaluates a completeness of the online learning. A determination unit determines whether or not the tracking unit is to use the first discriminator to track the subject according to the evaluation of the completeness.