Facial Recognition Aging Error Mitigation via Temporal Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current facial recognition technologies face inaccuracies due to face aging errors, especially when a large time gap exists between image capture dates or when dealing with children, as they rely on unreliable age simulation or transformation models.
Innovation Solution
A method that organizes images by timestamps and uses faces from successive time intervals as reference sets for facial recognition, incrementally searching and updating reference sets to improve matching accuracy, while also utilizing negative reference sets to exclude non-matching faces.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Duration of action of stationary object
If age simulation models or age transformation models are used to create aged models of faces for facial recognition, then facial recognition can be performed across different time periods, but the accuracy deteriorates when large time gaps exist or when dealing with children whose faces change significantly
Solution Approach 1:
The patent divides the image set into multiple time intervals and performs facial recognition incrementally across these intervals. Instead of directly comparing a reference face to all images regardless of time, the system segments the search space temporally and updates reference sets at each interval, thereby maintaining accuracy while covering large time spans.
Solution Approach 2:
The patent introduces a temporal dimension to the facial recognition process by organizing images into time intervals and using timestamps to guide the search. This dimensional approach allows the system to navigate through time systematically, updating reference faces at appropriate intervals to account for aging while maintaining recognition accuracy.
2Productivity
If a single reference face is used for searching across all images in the image set, then the search process is simple and fast, but the accuracy deteriorates when faces change significantly over time
Solution Approach 1:
The patent implements dynamic reference sets that are updated incrementally as the search progresses through time intervals. Instead of using a static reference face, the system dynamically adapts the reference set by incorporating faces found in each time interval, allowing it to track facial changes over time while maintaining search efficiency.
Solution Approach 2:
The system uses feedback from each time interval's search results to update the reference set for subsequent intervals. Faces identified in earlier intervals feed into the reference set for later intervals, creating a feedback loop that continuously refines the recognition accuracy while preserving search efficiency through incremental updates.
3Device complexity
If facial recognition is performed on all images simultaneously using a fixed reference set, then the implementation is straightforward, but the reliability deteriorates when dealing with large time gaps or significant facial changes
Solution Approach 1:
The patent segments the facial recognition process into multiple stages corresponding to different time intervals. This segmentation reduces the complexity of each individual recognition task while improving overall reliability by accounting for temporal changes in facial appearance through incremental reference set updates.
Data Source
AI summary
A computer implemented method and apparatus for mitigating face aging errors when performing facial recognition. The method comprises receiving an indication of a face that needs to be searched in an image set, where each image in the image set comprises a timestamp that identifies a creation date of the image, the creation date being in a continuum of successive time intervals; and identifying the indicated face in images taken in each time interval of a plurality of successive time intervals for the indicated face, wherein each face found in images taken in a previous successive time interval is used as a reference set for identifying the face in images taken in a next successive time interval.