Trajectory Inspection With Confidence Prediction for Similar Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing identification systems for foreign objects in liquids lack reliability in distinguishing between different types of objects based on time-series data, particularly in critical applications like liquid pharmaceuticals, and struggle to accurately predict the confidence levels of identification results.
Innovation Solution
An inspection system that includes an identification model trained using time-series data of object movement trajectories and a confidence level prediction model, utilizing observation specifications to differentiate between similar trajectories and predict the confidence levels of identification results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a confidence level prediction model is learned using only middle feature values from the identification model as training data, then the model can predict confidence levels, but it cannot differentiate between confidence levels of results estimated from similar input images or time-series data
Solution Approach 1:
The patent introduces a new dimension to the training data by incorporating observation specifications (such as observation time, observation position, and observation conditions) alongside the middle feature values. This additional dimensional information enables the confidence level prediction model to differentiate between similar input images and time-series data that would otherwise produce identical confidence predictions.
Solution Approach 2:
The patent changes the parameters used for training the confidence level prediction model by including not only the middle feature values but also observation specifications as training data. This parameter expansion allows the model to capture subtle differences in observation conditions and differentiate between similar trajectories more effectively.
2Productivity
If an identification model is used to determine object types based on movement trajectories, then identification can be performed, but the results may be wrong and reliability cannot be ensured in critical applications
Solution Approach 1:
The patent implements a feedback mechanism where the confidence level prediction model provides reliability information about the identification results. By analyzing the predicted confidence levels and observation specifications, the system can identify uncertain cases and trigger re-inspection or alternative verification processes, thereby ensuring reliability in critical applications while maintaining efficient identification for high-confidence cases.
Data Source
AI summary
An inspection system includes: an identification model learning means that performs machine-learning of a model identifying the type of a target object from time-series data representing the movement trajectory of the target object obtained by observation; a confidence level prediction model learning means that performs machine-learning of a confidence level prediction model estimating the confidence level of an estimation result by the identification model from the observation specification of time-series data representing the movement trajectory of a target object; and a determining means that uses the learned identification model to estimate the type of a target object from the movement trajectory of the target object obtained by observation, and uses the learned confidence level prediction model to predict the confidence level of an estimation result by the identification model from the observation specification of the time-series data.