On-Device Biometric Metadata Generation for Private Reinstallation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing applications on mobile devices incur time delays and processor expenditures by recalculating object identification upon reinstallation and expose biometric data to security risks by transferring it to external repositories.
Innovation Solution
A system that processes images on a mobile device to generate biometric data locally, using neural networks to identify faces, generate biometric reference maps, and store them securely on the device, ensuring privacy and efficiency by fetching preprocessed data during reinstallation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If biometric data is transferred to external repositories for analysis, then object identification can be performed, but security risks and privacy exposure increase
Solution Approach 1:
The patent extracts the biometric analysis function from external repositories and implements it locally on the mobile device. The neural network model is downloaded and executed on-device, allowing object identification to be performed without transferring biometric data to external servers, thus maintaining security while achieving accurate identification.
Solution Approach 2:
The patent introduces an on-device neural network model as an intermediary between the biometric data and external repositories. The model processes data locally and can share only aggregated or anonymized statistics with external servers, eliminating the need to transfer sensitive biometric information while maintaining identification capabilities.
2Reliability
If object identification is recalculated upon reinstallation, then accurate identification is achieved, but time delays and processor expenditures increase
Solution Approach 1:
The patent performs preliminary action by downloading the neural network model before reinstallation occurs. The model is stored locally on the device, so when the application is reinstalled, identification can resume immediately using the pre-loaded model without requiring recalculation or reprocessing of biometric data.
Solution Approach 2:
The patent creates a local copy of the neural network model on the mobile device. This copy allows the system to perform identification operations without accessing external resources or recalculating from scratch, significantly reducing processing time upon reinstallation while maintaining identification accuracy.
3Reliability
If biometric data is transferred to external resources for analysis, then object identification can be performed, but processor expenditures on the device are reduced
Solution Approach 1:
The patent extracts the computationally intensive neural network model from external servers and places it locally on the mobile device. This allows the device to perform all identification processing locally without continuous data transfer to external resources, reducing network energy consumption while maintaining processing capabilities.
Solution Approach 2:
The patent performs preliminary action by downloading and caching the neural network model before it is needed. The model is stored locally in advance, eliminating the need for repeated downloads or cloud processing during identification operations, thus reducing overall processor and network energy consumption.
Data Source
AI summary
Systems, devices, media and methods are presented for generating biometric image data. In one example, a system accesses a set of images stored on a mobile computing device. The system identifies one or more faces depicted in the set of images and generates a set of face images from the set of images. The system determines a set of positions of a set of facial features depicted within the set of face images and generates a set of biometric reference maps based on the set of positions. The system transmits the set of face images to a reference server and stores the set of biometric reference maps on the mobile computing device.