Trainingsprogramm, Trainingsverfahren und Trainingsvorrichtung

EP4790668 10. April 2025

Anmelder: FUJITSU LIMITED 🇯🇵

Details

Veröffentlichungs-Nr.
EP4790668
Anmeldetag
6. Oktober 2023
Veröffentlichung
10. April 2025
Rechtsraum
EP
IPC
G06V40/10
Offizieller Volltext

Abstract

To generate a neural network that enables biometric authentication robust to posture variations of a biological body part. A processing unit (12) converts a learning image (2-1), which is for machine learning of a neural network (1) configured to calculate authentication data for biometric authentication from an input biometric image, into three-dimensional data (3a) by using a first conversion expression based on internal parameters (13) of an imaging apparatus for capturing the biometric image; generates three-dimensional data (3b) by varying a posture of a biological body part (4) included in the three-dimensional data (3a); generates a learning image (2-1a) by converting the three-dimensional data (3b) by using a second conversion expression that performs inverse conversion of conversion performed by using the first conversion expression; and executes machine learning of the neural network (1) by using at least the learning image (2-1) and the learning image (2-1a).

Anmelder

Firma
FUJITSU LIMITED
Land
🇯🇵 Japan
🇯🇵 Fujitsu

Japanisches Unternehmen für Informationstechnik, das Computer, Server, Netzwerktechnik sowie IT-Dienstleistungen und Softwarelösungen entwickelt und anbietet.

17.908 Patente in unserer Datenbank

Vertreten von

Noch Fragen?

Wir helfen Ihnen gerne weiter. Schreiben Sie uns einfach.

Kontakt aufnehmen