Detection Model Memory Feature Updating for Fast Scene Migration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In manufacturing processes, scene migration of training models often requires generating sufficient sample data quickly, which is time-consuming and labor-intensive, especially for models deployed on edge devices that have limited computation power, and involves extensive data labeling and verification.
Innovation Solution
A detection method that updates a model's memory feature set by performing metric calculations on sample data from a target scene, generating a target memory feature set to replace the original set, allowing the model to adapt to the new scene with minimal data and computational effort.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional supervised training is used for model scene migration, then model accuracy can be maintained, but time consumption and labor intensity increase significantly due to extensive data labeling and verification
Solution Approach 1:
The patent extracts and updates only the memory feature set from the original training model, separating this critical component from the rest of the model structure. This allows selective updating of scene-specific features without retraining the entire model, significantly reducing time consumption while maintaining detection accuracy.
Solution Approach 2:
The patent performs preliminary extraction of memory feature sets from source scene data before migration. These pre-extracted features are then directly applicable to the target scene, eliminating the need for time-consuming iterative training and reducing overall migration time.
2Measurement precision
If traditional supervised training is used for model scene migration, then model accuracy can be maintained, but labor intensity increases due to extensive data labeling requirements
Solution Approach 1:
The patent extracts memory feature sets directly from source scene data without requiring manual labeling in the target scene. This extraction approach automates the feature acquisition process while maintaining accuracy, reducing labor intensity associated with data labeling.
Solution Approach 2:
The patent copies the memory feature set structure from the source scene model and adapts it to the target scene through metric-based updates. This copying mechanism eliminates the need for manual labeling while preserving the essential feature representations needed for accurate detection.
3Measurement precision
If iterative training processes are used for model adaptation, then model accuracy improves, but computational resources and training time are excessively consumed
Solution Approach 1:
The patent extracts and updates only the memory feature set rather than performing iterative training on the entire model. This selective extraction approach maintains adaptation accuracy while dramatically reducing computational resource consumption by focusing only on the essential scene-specific features.
Solution Approach 2:
The patent changes the approach from iterative parameter optimization to direct metric-based feature updates. By using distance metrics to guide feature set updates, the system achieves accurate adaptation without the computational overhead of multiple training iterations.
4Measurement precision
If sufficient sample data is generated for model training in a new scene, then model accuracy is maintained, but the process becomes time-consuming and labor-intensive
Solution Approach 1:
The patent extracts memory feature sets that capture essential scene characteristics without requiring large volumes of sample data. This extraction approach maintains model accuracy by preserving critical features while improving migration efficiency by eliminating the need for extensive data collection and labeling.
Solution Approach 2:
The patent uses a partial approach by updating only the memory feature set rather than retraining the entire model with extensive data. This partial action achieves sufficient accuracy for scene migration while dramatically improving productivity by reducing the data and time requirements.
Data Source
AI summary
A detection method includes obtaining a to-be-migrated model. The to-be-migrated model includes a memory feature set, and the memory feature set represents a feature vector set associated with an application scene corresponding to the to-be-migrated model. The method further includes performing a metric calculation on at least one piece of sample data of a target scene and the memory feature set to obtain at least one metric calculation result, and updating the memory feature set according to the at least one metric calculation result to obtain a target memory feature set. The target memory feature set represents a feature vector set associated with the target scene. The method further includes obtaining a target detection model by replacing the memory feature set of the to-be-migrated model with the target memory feature set.