US 12,711,803 B1Granted

System for biometric input assessment

AssigneeAmazon Technologies, Inc.

InventorsSaheb Chhabra, Rui Zhao, Manoj Aggarwal, Gerard Guy Medioni, Vishal Kaushal

Life of this patent

Estimating the term…

Loading patent…

Abstract

A machine learning network is trained to determine quality metrics of biometric input data. The quality metrics may include an estimated minutiae density, estimated feature point density, and so forth. Quality metrics are compared to threshold values to determine if the biometric input data is suitable for use. Different thresholds may be specified for enrollment and identification. A multi-branch machine learning network architecture may be used, with a first portion determining embedding data, and additional portions determining various quality metrics, such as the estimated minutiae density and estimated feature point density. This architecture substantially reduces latency in determining quality metrics. Biometric input data processed by the network may comprise an entire image, or patches of the image. Data associated with one or more of different images or different patches may be aggregated to determine embedding data used for identification. Aggregation may be based on the quality metrics.

Claims20

Dependent claims

A system comprising:

one or more memories, storing computer-executable instructions;
and
one or more hardware processors to execute the computer-executable instructions to:
determine first image data acquired by an input device at a first time, wherein the first image data comprises image data associated with a palm of a user;
determine first embedding data based on processing the first image data using a first portion of a machine learning network;
determine a first set of metric data based on processing the first image data using a second portion of the machine learning network, wherein the first set of metric data comprises:
a first value indicative of density of a first type of features, and
a second value indicative of density of a second type of features;
determine identification data;
determine that the first value is greater than a first threshold value;
determine that the second value is greater than a second threshold value;
and
based on the determination that the first value is greater than the first threshold value and the determination that the second value is greater than the second threshold value, store enrolled user data indicative of an association between the first embedding data and the identification data.

10 more elements

A computer-implemented method comprising:

determining first image data acquired by an input device at a first time, wherein the first image data comprises image data associated with a palm of a user;
determining first embedding data based on processing the first image data using a first portion of a machine learning network;
determining a first set of metric data based on processing the first image data using a second portion of the machine learning network, wherein the first set of metric data comprises:
a first value indicative of density of a first type of features, and
a second value indicative of density of a second type of features;
determining that the first value and the second value in the first set of metric data are greater than first threshold conditions;
and
based on the determination that the first value and the second value in the first set of metric data are greater than the first threshold conditions, storing enrolled user data indicative of an association between the first embedding data and identification data associated with the user.

5 more elements

A system comprising:

one or more memories, storing computer-executable instructions;
and
one or more hardware processors to execute the computer-executable instructions to:
determine first image data comprising a plurality of modalities acquired at a first time;
determine first embedding data based on the first image data;
determine a first set of metric data based on the first image data, wherein the first set of metric data comprises:
a first value indicative of density of a first type of features, and
a second value indicative of density of a second type of features;
determine that the first value and the second value in the first set of metric data are greater than first threshold conditions;
and
responsive to the determination that the first value and the second value in the first set of metric data are greater than the first threshold conditions, store enrolled user data indicative of an association between the first embedding data and identification data associated with a user.

8 more elements

Dependencies are inferred from the claim text.

Facts

Application: 18/337,936Filed 20 Jun 2023
Granted: 18 Aug 2026
Term ends: Not estimatedThe source records don’t support an estimate.

Classification6 codes

G06V 40/1353
Preprocessing; Feature extraction: extracting features related to minutiae or pores
G06V 10/7715
Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or…:…
G06V 10/774
Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or…:…
G06V 10/82
Using pattern recognition or machine learning: using neural networks