Face Image Grouping via Non-Negative Matrix Factorization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional face detection methods require prior training with known faces and supervision, making them inconvenient and less accurate for categorizing face images of different individuals.
Innovation Solution
A computer-implemented method that automatically categorizes face images using statistical analysis and non-negative matrix factorization, allowing for probabilistic grouping without prior retraining, and is compatible with various similarity functions and face descriptors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional face detection methods are used with prior training and supervision, then face recognition accuracy is improved, but system complexity and training time increase
Solution Approach 1:
The system performs self-service by automatically learning face groupings from unlabeled images without requiring external supervision or manual training data. The algorithm independently identifies face groups through iterative clustering and similarity analysis, eliminating the need for human-labeled training sets while maintaining high recognition accuracy
Solution Approach 2:
The system performs preliminary actions by pre-computing face embeddings and similarity matrices from the entire image set before actual grouping occurs. This preprocessing step creates reusable face representations that accelerate the subsequent iterative grouping process, reducing overall computational complexity
2Measurement precision
If prior training with known faces is required, then categorization accuracy is improved, but processing time and computational resources increase
Solution Approach 1:
The system eliminates the need for prior training with known faces by using self-service unsupervised learning. It automatically discovers face groupings through similarity-based clustering algorithms that work directly on the input image data, achieving high categorization accuracy without any training phase
Solution Approach 2:
The system changes the approach from supervised learning with fixed training parameters to unsupervised learning with adaptive parameters. The similarity threshold and clustering parameters are dynamically adjusted based on the data distribution, allowing the system to achieve high accuracy without predetermined training data
3Measurement precision
If supervised training methods are used, then face grouping accuracy is improved, but ease of operation deteriorates
Solution Approach 1:
The system provides self-service by automatically performing face grouping without requiring user supervision, manual annotation, or interactive training. Users simply input images and the system autonomously completes the entire grouping process, dramatically improving ease of operation while maintaining high accuracy
4Adaptability or versatility
If conventional face detection methods are used, then face detection capability is achieved, but adaptability to different scenarios is reduced
Solution Approach 1:
The system achieves universality by designing a flexible framework that can handle various face grouping scenarios including single-person detection, multiple persons, and group photos. The similarity-based approach and adjustable thresholds make it adaptable to different image types while maintaining reliable grouping accuracy across all scenarios
Data Source
AI summary
A computer-implemented method for sorting face images of different individuals into different groups includes obtaining face images comprising faces of unknown individuals by a computer processor; calculating similarity functions between pairs of face images by the computer processor; joining face images that have values of the similarity functions above a predetermined threshold into a hypothetical face group, wherein the face images in the hypothetical face group hypothetically belong to a same person; conducting non-negative matrix factorization on values of the similarity functions in the hypothetical face group to test truthfulness of the hypothetical face group; and identifying the hypothetical face group as a true face group if a percentage of the associated similarity functions being true is above a threshold based on the non-negative matrix factorization.