Facial Recognition Predictive Model for Intrapersonal Variation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional facial recognition techniques face challenges in accurately matching images due to intrapersonal variations such as different viewing angles, poses, illumination levels, facial expressions, and aging, which limit their effectiveness.
Innovation Solution
A predictive model is employed that uses an identity data set with multiple images of a person in various settings to predict the appearance of an input image under different conditions, allowing for appearance-prediction or likelihood-prediction approaches to recognize faces, and includes a switching mechanism to switch to direct appearance-matching when intrapersonal settings are similar.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional appearance-based facial recognition is used, then the system is simple and easy to implement, but it fails to accurately match images with large intrapersonal variations
Solution Approach 1:
The patent introduces an identity data set as an intermediary between the captured image and the stored images. This identity data set contains multiple images of the same person under different conditions (different poses, illuminations, expressions). The system uses this intermediary to bridge the gap caused by intrapersonal variations, allowing accurate matching even when the captured image differs significantly from stored images.
Solution Approach 2:
The system performs preliminary association of the captured image with the identity data set before performing the actual matching. This preliminary action involves identifying potential matches in the identity data set that account for intrapersonal variations, preparing the comparison in advance to handle different viewing angles, poses, and lighting conditions.
2Adaptability or versatility
If the system uses a predictive model with identity data set, then it can handle intrapersonal variations, but the processing time and computational complexity increase
Solution Approach 1:
The system applies partial action by selectively using the predictive model only when intrapersonal variations are detected. The switching mechanism evaluates whether the captured image exhibits significant variations from stored images, and only then activates the more computationally intensive predictive model with the identity data set. For images with minimal variations, the system uses the simpler conventional matching approach.
Solution Approach 2:
The system dynamically switches between two recognition approaches based on the characteristics of the input image. The switching mechanism adapts the processing approach in real-time: using conventional appearance-based matching for images with small intrapersonal variations, and switching to the predictive model with identity data set for images with large variations, thereby optimizing processing time.
3Productivity
If direct appearance-matching is used for all cases, then the processing is fast, but it fails when intrapersonal settings are very different
Solution Approach 1:
The system dynamically adjusts the recognition approach based on the degree of intrapersonal variation detected in the captured image. When variations are minimal, it uses fast direct appearance-matching. When variations are significant, it switches to the more reliable predictive model using the identity data set, thus maintaining both speed and reliability adaptively.
Solution Approach 2:
The switching mechanism provides feedback control by evaluating the characteristics of the captured image and selecting the appropriate recognition approach. It assesses whether intrapersonal variations are present and uses this feedback to determine whether to use the fast but less reliable direct matching or the slower but more reliable predictive model, ensuring optimal performance for each case.
Data Source
AI summary
Some implementations provide techniques and arrangements to address intrapersonal variations encountered during facial recognition. For example, some implementations employ an identity data set having a plurality of images representing different intrapersonal settings. A predictive model may associate one or more input images with one or more images in the identity data set. Some implementations may use an appearance-prediction approach to compare two images by predicting an appearance of at least one of the images under an intrapersonal setting of the other image. Further, some implementations may utilize a likelihood-prediction approach for comparing images that generates a classifier for an input image based on an association of an input image with the identity data set.