Facial Embedding Kin Verification for Age-Robust Family Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems struggle to identify children who have grown up and no longer resemble their childhood photos, complicating efforts to locate missing children and other family members, and methods like DNA testing are costly and inefficient.
Innovation Solution
A system using facial recognition embeddings and machine learning models to estimate kinship based on genetic similarity in facial features, trained with triplets of facial images to determine relationships between subjects and potential family members.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If DNA testing is used to verify family relationships, then accuracy of relationship identification is improved, but cost and effort increase significantly
Solution Approach 1:
The patent replaces the mechanical/biological DNA testing system with an optical/image-based facial recognition system. Instead of using DNA extraction, laboratory processing, and genetic analysis, the system uses facial image capture, feature extraction, and machine learning-based kinship verification to achieve relationship identification with significantly reduced cost and effort while maintaining acceptable accuracy
Solution Approach 2:
The patent uses facial images as a copy or representation of genetic information. Rather than directly analyzing DNA, the system captures visual representations (facial photos) that contain phenotypic information about genetic relationships, then processes these image copies through embedding models and similarity calculations to verify kinship
2Device complexity
If visual identification methods are used for grown children, then cost and effort are reduced, but accuracy decreases because they no longer resemble childhood photos
Solution Approach 1:
The system performs preliminary action by collecting and storing facial images of family members (including childhood photos when available) before the identification task. The embedding models are pre-trained on large datasets of facial images to learn age-invariant features. This preliminary preparation enables the system to handle the challenge of appearance changes over time by having pre-processed reference data and trained models ready for comparison
Solution Approach 2:
The patent changes the parameters used for comparison from direct pixel-based image matching to high-dimensional embedding space representations. The system transforms facial images into embedding vectors that capture essential genetic and phenotypic features, then performs similarity calculations in this transformed parameter space. This parameter transformation makes the comparison more robust to appearance changes due to aging
3Productivity
If simple facial comparison is used, then speed and ease of operation are improved, but reliability of kinship determination worsens
Solution Approach 1:
The patent moves the comparison from two-dimensional image space to high-dimensional embedding space. Instead of comparing pixels or simple facial features directly, the system transforms images into hundreds or thousands of dimensional embedding vectors that capture complex genetic information. This dimensional transformation enables more reliable kinship determination while maintaining computational efficiency through vector similarity calculations
Data Source
AI summary
For a given pair of biological parents and a possible child, the present system(s), method(s), and/or software estimate whether the child is indeed a biological child of the biological parents. Using a face recognition engine to determine embeddings from facial images, genetic similarity expressed in facial features is used to estimate kinship. Knowing that a child carries 50% of the genes from each parent, for example, and using facial features from both parents, chances for the present system(s), method(s), and or software to determine a correct match are increased considerably compared to prior systems.