WO2021213797A1 - Procédé de protection de données personnelles - Google Patents

Procédé de protection de données personnelles Download PDF

Info

Publication number
WO2021213797A1
WO2021213797A1 PCT/EP2021/058910 EP2021058910W WO2021213797A1 WO 2021213797 A1 WO2021213797 A1 WO 2021213797A1 EP 2021058910 W EP2021058910 W EP 2021058910W WO 2021213797 A1 WO2021213797 A1 WO 2021213797A1
Authority
WO
WIPO (PCT)
Prior art keywords
vector
facial features
neural network
personal data
identified facial
Prior art date
Application number
PCT/EP2021/058910
Other languages
German (de)
English (en)
Inventor
Johannes RADMER
Original Assignee
HELLA GmbH & Co. KGaA
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by HELLA GmbH & Co. KGaA filed Critical HELLA GmbH & Co. KGaA
Publication of WO2021213797A1 publication Critical patent/WO2021213797A1/fr

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/161—Detection; Localisation; Normalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects

Definitions

  • the invention relates to a method for protecting personal data in a surveillance system, such as, for example, in a shop, e.g. B. a shop, a restaurant, a hospital, in an airport and / or in a stadium, according to the independent method claim.
  • the invention also relates to a corresponding monitoring system according to the independent device claim.
  • the invention also relates to a corresponding computer program product for performing the method according to the invention.
  • Monitoring systems are known in principle from the prior art.
  • Known surveillance systems mostly have one or more camera units in order to detect and possibly recognize customers in a shop.
  • Customers can be classified according to properties, such as: B. Age, gender, etc.
  • the properties are extracted from features from one or more camera images.
  • the features are usually facial features.
  • the features are extracted using a machine learning process.
  • the camera can preferably be installed near an entrance to the shop. For example, a customer entering a store can be captured by the camera at the entrance. It can also be determined from the image whether the customer is female or male and / or what age he is.
  • the data or information obtained in this way can be used, for example, to determine the behavior of customers in general or of a specific customer in the shop (see, for example, EP 2808838 A1).
  • the object of the invention is therefore to provide an improved method for protecting personal data in a monitoring system, such as, for example, in a shop, e.g. B. a shop, a restaurant, a hospital, in an airport and / or in a stadium, which can at least partially overcome at least one of the above-mentioned problems.
  • a shop e.g. B. a shop, a restaurant, a hospital
  • an airport and / or in a stadium which can at least partially overcome at least one of the above-mentioned problems.
  • the object according to the invention is achieved by a method for protecting personal data in a monitoring system, such as, for example, in a shop, e.g. B. a shop, a restaurant, a hospital, in an airport and / or in a stadium, solved with the features of the independent method claim.
  • a monitoring system such as, for example, in a shop, e.g. B. a shop, a restaurant, a hospital, in an airport and / or in a stadium
  • the object according to the invention is achieved by a corresponding monitoring system according to the independent device claim and a corresponding computer program product.
  • the invention provides a method for protecting personal data in a monitoring system, such as, for example, in a shop, such as, for example, a shop, a restaurant, a hospital, in an airport and / or in a stadium, comprising the following steps :
  • Creating a re-identification vector (which can act as a key) for the vector (e.g. from the vector or with the aid of the vector or on the basis of the vector) of identified facial features with the aid of the second neural network in order to convert the vector of identified facial features anonymize (or encrypt and thus reliably protect against attacks by third parties).
  • a recording device such.
  • a camera unit for example a mono camera or a stereo camera, and / or a people counter device, initially takes one or more pictures.
  • the camera unit can be arranged at the entrance to a shop, for example.
  • the recorded image or images are processed with the aid of a first neural network (for example a convolutional neural network).
  • the first neural network recognizes the faces or identifies facial features in the image or in the images.
  • the first neural network can be located in a processor outside the camera or in the camera itself.
  • the first neural network generates a vector with identified facial features.
  • the first neural network can also generate personal data or information (age, gender, etc.) from the facial features. A set of personal data or information can advantageously be generated for each recognized face or for each vector of identified facial features.
  • the vectors with identified facial features are sent to a second neural network.
  • the second neural network can also be implemented, for example, in the form of a folding neural network.
  • the second neural network calculates (according to a certain principle, it is conceivable that the key may or may not be deducible from the vector of identified facial features) a so-called re-identification vector or key for each vector of identified facial features or in short expressed every facial feature vector that comes from the first neural network.
  • the re-identification vectors can, for example, be vectors or matrices that contain several lines, e.g. B. 2 to 200, 50 to 180, preferably 180 rows, and / or several columns, e.g. B.
  • each row and / or each column can have a number or a facial feature.
  • Each re-identification vector can thus be linked to a specific facial feature vector.
  • the re-identification vectors can advantageously serve as a key for the personal data, in particular to protect the personal data from external interference by unauthorized third parties.
  • the re-identification vectors within the meaning of the invention are advantageously pure “keys” since they do not contain any information about facial features or personal data.
  • the re-identification vector can advantageously be permuted at random after a certain time interval, for example every N hours.
  • the old key or the old re-identification vector can then be irrevocably deleted.
  • the time interval within which a re-identification vector is active can be set manually. In a preferred example, this interval can be determined as the opening time of the store.
  • the invention ensures that the re-identification vectors can only be used to identify the set of personal data within the specified time interval. If unauthorized persons were to read the personal data (which are protected by means of the re-identification vectors), they would not be able to find any connection between data and persons.
  • the method is particularly suitable for use in a shop. However, it can be used in any area where personal data should be protected.
  • the invention can provide that in step
  • the at least one image is recorded by at least one recording device.
  • the at least one recording device can be designed in the form of a camera unit, for example a mono camera or a stereo camera, and / or a people counter device.
  • images can be recorded that can be processed by a first neural network in order to identify facial features within the images and from the identified facial features personal data, such as. B. age, gender, etc. to be able to deduce.
  • the at least one recording device can have several devices of the same or different types. In this way, different evaluable data can be provided with the aid of the at least one recording device, which can enable easier, faster and / or more precise recognition of facial features.
  • the invention can provide that the step
  • the invention can provide that the step
  • a computing unit wherein the computing unit can be implemented in the at least one recording device that executes step 1), or wherein the computing unit can be implemented as a separate computing unit, which can be implemented by the at least one recording device is spaced apart.
  • a computing unit can thus be made available within the at least one recording device or a separate computing unit can be used, for example the user's own computing unit.
  • the method according to the invention can be provided in the form of a computer program product which, when it is played on the computing unit of the user, carries out a method within the meaning of the invention.
  • the invention can provide that the first neural network and / or the second neural network are / is provided in the form of a folding neural network. With the aid of such neural networks, the method according to the invention can be carried out with little working memory and computing power.
  • the invention can provide that in step
  • the invention can provide that the re-identification vector has a matrix form with several rows, for example 180 rows, and / or with several columns, and / or that at least one row and / or at least one column of the re-identification vector has a number, in particular a facial feature.
  • the re-identification vector has a matrix form with several rows, for example 180 rows, and / or with several columns, and / or that at least one row and / or at least one column of the re-identification vector has a number, in particular a facial feature.
  • the invention can provide that in step 4) the re-identification vector is derived from the vector of identified facial features.
  • the re-identification vector can be generated by permutating the vector of identified facial features.
  • the re-identification vector is permuted anew immediately after it has been created.
  • the vector of identified facial features can no longer be derived from the re-identification vector.
  • the invention can provide that the re-identification vector is generated randomly in step 4). There is no longer any connection with the vector of identified facial features.
  • One advantage here is that the data is protected even if the re-identification vector is saved.
  • the invention can advantageously provide that in step 4) the re-identification vector is permuted after a certain interval, in particular periodically and / or repeatedly. In this way, the security in the method according to the invention can be increased.
  • the invention can provide that the method has at least one of the following steps:
  • the user can decide and set when and how often the re-identification vector is permuted as desired and required.
  • security areas such as B. Hospitals and / or airports, for example. Longer intervals can be advantageous in order to be able to trace back to the person in the event of a violation.
  • the surveillance systems of daily life such as B. in a shop, it could be advantageous to permute the re-identification vector after each opening period.
  • the method has at least one further step:
  • a condition of use of the monitoring system in a shop could e.g. B. the opening time.
  • Another condition of use of the monitoring system can be, for example, the security level. The higher the security level, the longer the appropriate interval can be selected.
  • the invention can advantageously provide that the method has at least one further step:
  • the invention represents a monitoring system, such as for a shop, comprising: at least one recording device for recording at least one image that can depict a person, and a computing unit for analyzing the at least one image, the computing unit for this purpose is designed to carry out a method according to any one of the above claims.
  • the computing unit can advantageously have a memory in which the steps according to the invention are stored, for example in the form of a computer program product, the method according to the invention being able to be carried out when the steps according to the invention are carried out on the computing unit.
  • a computer program product can be provided which is stored in a portable and / or virtual memory and which carries out the method according to the invention when at least partially executed on a processing unit.
  • 1 shows an exemplary flow chart of a method according to the invention and 2 shows an exemplary representation of a monitoring system within the meaning of the invention.
  • FIG. 1 shows a schematic sequence of a method within the meaning of the invention that is used to protect personal data in a monitoring system 100, such as, for example, in a shop, such as, for example, a shop, a restaurant, a hospital, in an airport and / or in a stadium, serves.
  • the process has the following steps:
  • Creating a re-identification vector RV (which can act as a key) for the vector V1 (for example from the vector V1 or using the vector V1 or based on the vector V1) of identified facial features M using the second neural network N2 in order to anonymize the vector V1 of identified facial features M (or to encrypt it and thus reliably protect it from attacks by third parties).
  • a recording device G such as. B. a camera unit K, for example.
  • a camera unit K Comprising a mono camera and / or a stereo camera, and / or a people counter device PC, initially records one image B or several images B.
  • the camera unit K can be arranged, for example, at the entrance to a shop or in the course of the aisles in the shop.
  • the at least one recording device G has several devices of the same type or of different types.
  • the camera unit K can have several cameras which can be arranged in the course of the aisles in the shop.
  • the recorded image or images B are processed with the aid of a first neural network N1 (for example a convolutional neural network).
  • the first neural network N1 recognizes the faces or identifies facial features M in the image or images B.
  • the first neural network N1 can be provided in a computing unit R outside the camera unit K or in the camera unit K itself.
  • a vector V1 with identified facial features M is generated by the first neural network N1.
  • the first neural network N1 can also generate personal data D or information (age, gender, etc.) from the facial features M.
  • personal data D or information for each recognized face or for each vector V1 of identified facial features M, a set of personal data V 2 or information can advantageously be generated.
  • the vectors V1 with identified facial features M, and in particular no corresponding set V2 of personal data D, are sent to a second neural network N2 according to the invention.
  • the second neural network N2 can also be designed, for example, in the form of a folding neural network.
  • the second neural network N2 calculates (according to a certain principle, it is conceivable that the key can be derived from the vector V1 of identified facial features M or can be independent thereof) a so-called re-identification vector RV or key for each Vector V1 of identified facial features M, which comes from the first neural network N1.
  • the re-identification vectors RV can be, for example, vectors or matrices that contain several lines, e.g. B. 2 to 200, 50 to 180, preferably 180 rows, and / or several columns, e.g. B. 2 to 200, 50 to 180, preferably 180 columns, may include. It is conceivable that each row and / or each column can have a number or a facial feature M.
  • Each re-identification vector RV can thus be assigned to a specific facial feature vector V1.
  • the re-identification vectors RV can advantageously serve as a key for the personal data D in order, in particular, to protect the personal data D from external interference by unauthorized third parties.
  • the re-identification vectors RV can advantageously be pure “keys” and contain no information about the facial features M and / or about the personal data D.
  • the re-identification vector RV can advantageously be permuted at random after a specific time interval dt, for example every N hours.
  • the old key or the old re-identification vector RV can then be irrevocably deleted.
  • the re-identification vector RV can be derived from the vector V1 of identified facial features M, it can be advantageous for the re-identification vector RV to be permuted directly after it has been created.
  • the time interval dt within which a re-identification vector RV is active can be set manually and / or automatically. In a preferred example, this interval can be determined as the opening time of the store.
  • the invention ensures that the re-identification vectors can only be used for identifying the set V 2 of personal data D within the defined time interval dt. If unauthorized persons were to read the personal data D (which are protected by means of the re-identification vectors RV), they could not find any connection between data D and persons P or the faces G of persons P.
  • the method is particularly suitable for use in a shop, as FIG. 2 shows schematically.
  • the procedure can be used in any area where personal data should be protected.
  • the re-identification vector RV from the vector V1 of identified facial features M can be derived.
  • the re-identification vector RV can be generated by permutation of the vector V1 of identified facial features M or by multiplication with a permutation matrix.
  • the re-identification vector RV is permuted anew immediately after it has been created.
  • the vector V1 of identified facial features M can no longer be derived from the re-identification vector RV.
  • the re-identification vector RV can be generated randomly in step 4).
  • the connection to the vector V1 of identified facial features M can thus be deleted.
  • One advantage here is that the data D are also protected when the re-identification vector RV is stored for a period of time, for example for a time interval dt.
  • the method can have at least one of the following steps:
  • the method can have at least one further step:
  • the method can have at least one further step: 7) Providing a confirmation option for the calculated interval dt for a user of the monitoring system 100. This can also be done via a display DP.
  • a corresponding monitoring system 100 which is shown by way of example in FIG. 2, for example for a shop, also represents an aspect of the invention P can map, and a computing unit R for analyzing the at least one image B, wherein the computing unit R is designed to carry out a method according to one of the above claims.
  • the computing unit R can, for example, have a memory in which the steps according to the invention, e.g. B. in the form of a computer program product, can be stored, the inventive method can be carried out when executing the inventive steps on the computing unit R.
  • a third aspect of the invention can provide a computer program product which can be stored in a portable and / or virtual memory and which, when at least partially executed on a computing unit R, can carry out the method according to the invention according to the steps according to the invention.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Multimedia (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Computational Linguistics (AREA)
  • Molecular Biology (AREA)
  • Biophysics (AREA)
  • General Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Human Computer Interaction (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)
  • Collating Specific Patterns (AREA)

Abstract

L'invention concerne un procédé de protection de données personnelles (D) dans un système de surveillance (100), comportant les étapes suivantes : 1) enregistrer au moins une image (B), qui peut représenter une personne (P), 2) analyser la ou les images (B) en utilisant un premier réseau neuronal (N1) afin d'identifier des caractéristiques faciales (M) dans l'image (B) et de dériver des données personnelles (D), par exemple l'âge, le sexe, etc., à partir des caractéristiques faciales identifiées, 3) fournir, en particulier transmettre, un vecteur (V1) de caractéristiques faciales identifiées (M) à un deuxième réseau neuronal (N2), 4) créer un vecteur de réidentification (RV) pour le vecteur (V1) des caractéristiques faciales identifiées (M) en utilisant le deuxième réseau neuronal (2), afin d'anonymiser (ou par exemple de chiffrer) le vecteur (V1) de caractéristiques faciales identifiées (M).
PCT/EP2021/058910 2020-04-23 2021-04-06 Procédé de protection de données personnelles WO2021213797A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
DE102020111045.6A DE102020111045A1 (de) 2020-04-23 2020-04-23 Verfahren zum Schützen von personenbezogenen Daten
DE102020111045.6 2020-04-23

Publications (1)

Publication Number Publication Date
WO2021213797A1 true WO2021213797A1 (fr) 2021-10-28

Family

ID=75478019

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2021/058910 WO2021213797A1 (fr) 2020-04-23 2021-04-06 Procédé de protection de données personnelles

Country Status (2)

Country Link
DE (1) DE102020111045A1 (fr)
WO (1) WO2021213797A1 (fr)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2808838A1 (fr) 2013-05-30 2014-12-03 Panasonic Corporation Dispositif d'analyse de catégorie de client, système d'analyse de catégorie de client et procédé d'analyse de catégorie de client
DE102017215283A1 (de) * 2017-08-31 2019-02-28 Audi Ag Verfahren zum Anonymisieren eines Bilds für ein Kamerasystem eines Kraftfahrzeugs, Bildverarbeitungseinrichtung, Kamerasystem sowie Kraftfahrzeug

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102014101495B4 (de) 2014-02-06 2019-06-19 Fujitsu Technology Solutions Intellectual Property Gmbh Verfahren zum Zugang zu einem physisch abgesicherten Rack sowie Computernetz-Infrastruktur
EP3704624A1 (fr) 2017-11-03 2020-09-09 Veridas Digital Authentication Solutions, S.L. Procédés et dispositifs de vérification biométrique
US11392802B2 (en) 2018-03-07 2022-07-19 Private Identity Llc Systems and methods for privacy-enabled biometric processing
CA3050478A1 (fr) 2018-07-24 2020-01-24 Edison U. Ortiz Systemes et methodes de justificatifs d`identite sous la forme de jetons securises

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2808838A1 (fr) 2013-05-30 2014-12-03 Panasonic Corporation Dispositif d'analyse de catégorie de client, système d'analyse de catégorie de client et procédé d'analyse de catégorie de client
DE102017215283A1 (de) * 2017-08-31 2019-02-28 Audi Ag Verfahren zum Anonymisieren eines Bilds für ein Kamerasystem eines Kraftfahrzeugs, Bildverarbeitungseinrichtung, Kamerasystem sowie Kraftfahrzeug

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
HUKKELÅS HÅKON ET AL: "DeepPrivacy: A Generative Adversarial Network for Face Anonymization", 21 October 2019, ADVANCES IN INTELLIGENT DATA ANALYSIS XIX; [LECTURE NOTES IN COMPUTER SCIENCE; LECT.NOTES COMPUTER], SPRINGER INTERNATIONAL PUBLISHING, CHAM, PAGE(S) 565 - 578, ISBN: 978-3-030-71592-2, ISSN: 0302-9743, XP047525299 *
JOHNSON JUSTIN ET AL: "Perceptual Losses for Real-Time Style Transfer and Super-Resolution", 17 September 2016, ICIAP: INTERNATIONAL CONFERENCE ON IMAGE ANALYSIS AND PROCESSING, 17TH INTERNATIONAL CONFERENCE, NAPLES, ITALY, SEPTEMBER 9-13, 2013. PROCEEDINGS; [LECTURE NOTES IN COMPUTER SCIENCE; LECT.NOTES COMPUTER], SPRINGER, BERLIN, HEIDELBERG, PAGE(S) 694 - 7, ISBN: 978-3-642-17318-9, XP047356366 *
REN ZHONGZHENG ET AL: "Learning to Anonymize Faces for Privacy Preserving Action Detection", 6 October 2018, ICIAP: INTERNATIONAL CONFERENCE ON IMAGE ANALYSIS AND PROCESSING, 17TH INTERNATIONAL CONFERENCE, NAPLES, ITALY, SEPTEMBER 9-13, 2013. PROCEEDINGS; [LECTURE NOTES IN COMPUTER SCIENCE; LECT.NOTES COMPUTER], SPRINGER, BERLIN, HEIDELBERG, PAGE(S) 639 - 6, ISBN: 978-3-642-17318-9, XP047488312 *
RIBARIC SLOBODAN ET AL: "De-identification for privacy protection in multimedia content: A survey", SIGNAL PROCESSING. IMAGE COMMUNICATION, ELSEVIER SCIENCE PUBLISHERS, AMSTERDAM, NL, vol. 47, 1 June 2016 (2016-06-01), pages 131 - 151, XP029753254, ISSN: 0923-5965, DOI: 10.1016/J.IMAGE.2016.05.020 *
ZHANG KAIPENG ET AL: "Joint Face Detection and Alignment Using Multitask Cascaded Convolutional Networks", IEEE SIGNAL PROCESSING LETTERS, IEEE SERVICE CENTER, PISCATAWAY, NJ, US, vol. 23, no. 10, 1 October 2016 (2016-10-01), pages 1499 - 1503, XP011622636, ISSN: 1070-9908, [retrieved on 20160909], DOI: 10.1109/LSP.2016.2603342 *

Also Published As

Publication number Publication date
DE102020111045A1 (de) 2021-10-28

Similar Documents

Publication Publication Date Title
DE69228315T2 (de) Zugangskontrollsysteme
WO2018229246A1 (fr) Procédé mis en œuvre par ordinateur et produit de programme informatique destinés au contrôle d'accè d'un terminal
DE102019109288A1 (de) Anlage und Verfahren für Analyse des importierten Videos
WO2000043960A1 (fr) Procede et systeme de protection contre la fraude en cas de reconnaissance biometrique de personnes
DE102009044083A9 (de) System und Verfahren zur Zählung der Personenanzahl
EP4121885A1 (fr) Dispositif d'anonymisation, appareil de surveillance, procédé, programme informatique et support d'informations
DE102007050024A1 (de) Personenkontrollsystem und Verfahren zum Durchführen einer Personenkontrolle
DE112022001468T5 (de) Selektives unkenntlichmachen von bildern
EP2505540A1 (fr) Dispositif de surveillance d'accès doté d'au moins une unité vidéo
DE102020115697A1 (de) Verfahren und Vorrichtung zur natürlichen Gesichtsanonymisierung in Echtzeit
WO2021213797A1 (fr) Procédé de protection de données personnelles
DE102014207439A1 (de) Maskierung von sensiblen Daten bei der Benutzer-Identifikation
DE102022211513A1 (de) System und Verfahren zum Verarbeiten einer Datensubjekt-Rechteanforderung unter Verwendung von biometrischem Datenabgleich
EP2462557B1 (fr) Procédé d'analyse vidéo
EP3723339B1 (fr) Libération sécurisée d'une fonction protégée
EP3531357A1 (fr) Système de caisse, procédé d'authentification sur un système de caisse, programme informatique ainsi que support lisible par ordinateur
DE102018222605A1 (de) Verfahren zur Auswertung thermografischer Bilder
EP4227839B1 (fr) Dispositif de détection d'heure et/ou de présence, système, procédé de fonctionnement d'un dispositif de détection d'heure et/ou de présence, produit-programme informatique
EP4258228A1 (fr) Procédé de détection faciale précise et de détection de manipulations d'image faciale
EP4336462A1 (fr) Procédé de détection faciale précise et de détection de manipulations d'image faciale
DE102022122596A1 (de) Verfahren zur präzisen Gesichtserkennung und Detektion von Gesichtsbildmanipulationen
DE102009018725A1 (de) Verschlüsselung von alphanumerischen Zeichen und gesicherte Eingabe von alphanumerischen Zugangskodes
WO2023186652A1 (fr) Procédé de fonctionnement d'un dispositif de surveillance, programme informatique, support de stockage et dispositif de surveillance
DE102022124561A1 (de) Verfahren zur präzisen Gesichtserkennung und Detektion von Gesichtsbildmanipulationen
EP4336471A1 (fr) Procédé de détection faciale précise et de détection de manipulations d'image faciale

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21718062

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21718062

Country of ref document: EP

Kind code of ref document: A1