DE102022103850A1 - Method for adapting a function of a household appliance, preferably a kitchen appliance - Google Patents

Abstract

The invention relates to a method for adapting a function of a household appliance (1), preferably a kitchen appliance (1), with at least the following steps:• optically detecting (100) an area adjoining the household appliance (1) as image data,• detecting (200) a hand (2) of a user in the optically captured image data,• determining (300) a feature vector of the recognized hand (2),• comparing (400) the determined feature vector of the recognized hand (2) with at least one predetermined feature - vector of a predetermined hand and• if the determined feature vector matches the predetermined feature vector, adapting (500) the function of the household appliance (1) as a function of the predetermined hand.

Description

The invention relates to a method for adapting a function of a household appliance, preferably a kitchen appliance, and a household appliance, a control unit and a computer program product for this.

It is known for a wide variety of electronic devices that they can have a number of functions or different settings, parameters and the like for at least one function. A user of the electronic device can thus choose between different functions or function settings in order to be able to use the electronic device in as many different ways as possible and according to their own ideas, preferences and needs.

These electronic devices also include household appliances, which, for example as a washing machine, can have different washing programs between which the user can choose and which the user can set or parameterize differently. This applies similarly to tumble dryers, washer-dryers, vacuum cleaners and the like.

Household appliances also include kitchen appliances, which can also be used or configured differently by the user. For example, different hotplates can be used on a hob and, for example, their power levels can be set by the user. This also applies to a cooktop extractor hood. Other such kitchen appliances can be, for example, microwaves, steam cookers, ovens and combi appliances, but also hot drinks machines, kettles, warming drawers and the like.

Such electronic devices can usually be used by several people as users, whose user behavior usually differs with regard to the functions used and their settings or parameters. Accordingly, devices of this type should always be switched on or made available in a general basic state, from which the different users can start using them individually.

The disadvantage here is that each user, starting from the general basic state of the electronic device, has to make certain and usually the same settings of functions each time it is used in order to adapt the electronic device to his individual use. For example, in a washing machine, the temperature and speed can be set after the washing program has been selected. In the case of a hot drinks machine, the hot drink can be selected and then the amount of beans, the water temperature and the like can be set, for example to prepare a coffee drink. In the case of an oven, the cooking time and the cooking temperature can be set after the program or function has been selected. These settings, which are always or usually the same and always have to be made, can cost the user time and annoy them.

It can therefore be desirable for user-specific information about the desired settings or parameters to be able to be made automatically by the electronic device depending on the respective user. In other words, such an electronic device should know the user-specific settings or functions and be able to make them automatically when the user in question uses or wants to use the electronic device.

For this purpose, it is generally known that an electronic device automatically recognizes the respective user of a predetermined number of users, with corresponding information regarding the use of the electronic device, such as functions and/or their settings, being specified or provided for each user in the electronic device. are deposited.

Such a user recognition can take place, for example, by means of known methods of image recognition, in that the user's face is recorded, for example, by means of a camera and recognized by means of image recognition.

The disadvantage here, however, is the corresponding effort involved in arranging and aligning a camera with respect to the electronic device in such a way that the user can be reliably optically recorded. The quality of the recorded image data must always be sufficient for the subsequent image recognition. This can be problematic or expensive if it is to be ensured that the face of each user can be reliably detected and recognized regardless of their body size and their position and orientation in relation to the electronic device. If this requires the user to consciously look directly into the camera, which can be arranged at a rather low height for this purpose, the user can find this uncomfortable and cumbersome, which can lead to the rejection of this possibility of user identification.

It is also disadvantageous that such a camera directed into the room, for example into the kitchen, can optically record this area throughout. This can lead to the possibility of continuous camera surveillance, for example of part of the kitchen, which the user also finds unpleasant and therefore the recognition of the user by means of facial recognition can be rejected as described above.

• • Einlesen eines Bildsignals über eine Schnittstelle zu einer Kamera zum Erfassen eines an das Haushaltgerät angrenzenden Bereiches, wobei das Bildsignal ein von der Kamera aufgenommenes Bild repräsentiert;
• • Bestimmen eines Handsignals unter Verwendung des Bildsignals, wobei das Handsignal eine Charakteristik einer in dem Bild abgebildeten Hand eines Nutzers des Haushaltgerätes repräsentiert; und
• • Bereitstellen eines Anpassungssignals zum Anpassen der Funktion des Haushaltgerätes unter Verwendung des Handsignals.

An internal method of the applicant for adapting a function of a household appliance, the method having the following steps:

• • reading in an image signal via an interface to a camera for capturing an area adjoining the household appliance, the image signal representing an image recorded by the camera;
• • determining a hand signal using the image signal, the hand signal representing a characteristic of a hand of a user of the household appliance depicted in the image; and
• • Providing an adjustment signal for adjusting the function of the household appliance using the hand signal.

The advantage here is that the area in which the electronic device is operated by the user with his or her hands is comparatively small and defined, so that the camera can be arranged and designed very specifically for this area optically capture the area in the required quality. This can keep the expense of the optical image acquisition and image recognition comparatively low and ensure implementation.

It is also advantageous that to capture this area, the camera can be aligned obliquely from above or from the side and can thus optically capture a comparatively small section of the area surrounding the electronic device, such as the kitchen. This can limit the camera's field of view or detection area sufficiently so that the camera cannot be used for room surveillance.

The invention therefore faces the problem of expanding and/or improving the possibilities for individualized use of electronic devices. This should be done in particular for household appliances and especially for kitchen appliances. In particular, the internally known procedure is to be improved. At least an alternative to the known possibilities should be created.

According to the invention, this problem is solved by a method, a household appliance, a control unit and a computer program product having the features of the independent claims. Advantageous configurations and developments of the invention result from the following dependent claims.

• • optisches Erfassen eines an das Haushaltegerät angrenzenden Bereichs als Bilddaten,
• • Erkennen einer Hand oder zweier Hände eines Benutzers in den optisch erfassten Bilddaten,
• • Bestimmen eines Feature-Vektors der erkannten Hand oder der Hände,
• • Vergleichen des bestimmten Feature-Vektors der erkannten Hand oder der Hände mit wenigstens einem vorbestimmten Feature-Vektor einer vorbestimmten Hand oder zweier Hände und
• • bei Übereinstimmung des bestimmten Feature-Vektors mit dem vorbestimmten Feature-Vektor, Anpassen der Funktion des Haushaltsgeräts in Abhängigkeit der vorbestimmten Hand.

The present invention thus relates to a method for adapting a function of a household appliance, preferably a kitchen appliance, having at least the following steps:

• • Optical detection of an area adjacent to the household appliance as image data,
• • Recognizing a hand or two hands of a user in the optically captured image data,
• • determining a feature vector of the detected hand or hands,
• • comparing the determined feature vector of the recognized hand or hands with at least one predetermined feature vector of a predetermined hand or two hands and
• • if the determined feature vector matches the predetermined feature vector, adapting the function of the household appliance depending on the predetermined hand.

Such household appliances or kitchen appliances can in particular be electronic appliances which can be operated or adjusted by a person as the user with at least one hand in order to be used as intended. The user can select a function of the household appliance with his hand on an input element or control element and set or parameterize its specific implementation in order to be able to adapt the function of the household appliance to his personal and individual wishes or needs.

According to the invention, the user can be relieved of having to perform the previously described setting or parameterization of the function each time the household appliance is used, by the hand of the user or the user being recognized by his hand by the household appliance and the function of the household appliance corresponding to the recognized Hand or the recognized user can be automatically parameterized or set by the household appliance or its control unit. This can relieve the user of this, which can save time and effort for the user and thus increase the convenience of using the household appliance.

This can be done according to the invention that the area of the household appliance in wel chem the operation of the household appliance takes place by the user with his hand, optically recorded by means of an optical image acquisition unit such as an optical camera and these images are used or temporarily stored as image data. Using known methods of image processing and pattern recognition, a human hand can be recognized in the recorded image data, which can be regarded as the hand of the user who is currently using or operating the household appliance.

A feature vector of the recognized hand can now be determined in the optically recorded image data. A feature vector or a feature vector is understood to be a vector which summarizes the (numerically) parameterizable properties of a pattern in a vector way. Different features characteristic of the pattern make up the different dimensions of this vector. The totality of possible feature vectors is called the feature space. Feature vectors facilitate automatic classification because they greatly reduce the properties to be classified, so that instead of a complete image, for example, only a vector of ten numbers needs to be considered. This can reduce the effort of data processing and speed up data processing.

The specific feature vector of the recognized hand can then be compared with at least one predetermined feature vector of a predetermined hand. The predetermined feature vector belongs to a predetermined hand of a predetermined user and may have been determined beforehand by means of method steps of predetermining, as will be explained in more detail below. A predetermined parameterization of the function of the household appliance can be assigned to the predetermined hand, which corresponds to the usual or desired parameterization of this function of the household appliance of the user of the predetermined hand. The predetermined hand can also be assigned a unique identification of the user of the predetermined hand, in which case the user's usual or desired parameterization of this function of the household appliance can then be assigned to the unique identification of the user. In any case, an individual setting or adaptation of this function of the household appliance, which corresponds to a predetermined user, can be assigned to the predetermined feature vector.

If the comparison now results in an at least adequate match of the specific feature vector with the predetermined feature vector, the function of the household appliance can be adapted in response thereto as a function of the predetermined hand. As a result, the function of the household appliance can be adapted to the current user, since the user's recognized hand can be used to assign the function to predetermined or stored adaptations or parameterization. This can relieve the user of having to make the appropriate settings and still enable the household appliance to be used in accordance with the customary or desired specifications of the individual user.

Assigning the optically detected hand of the current user to at least one setting or adjustment of at least one predetermined stored hand using feature vectors or feature vectors can reduce the corresponding effort, accelerate the procedure and/or improve the quality of the result of the comparison increase.

According to one aspect of the invention, at least one parameterization of at least one function of the household appliance is assigned to the predetermined hand. As already mentioned, this can represent a possibility for the corresponding assignment, which can be simpler or faster than the additional assignment of a unique identification of the current user, as will be described below.

According to a further aspect of the invention, a predetermined user is assigned to the predetermined hand by means of a unique identification, and at least one parameterization of at least one function of the household appliance is assigned to the predetermined user. As already mentioned, this can represent an alternative possibility for the corresponding assignment. Here, the user can be recognized personally, which can enable further usage options. For example, the current user can be addressed by name via a display element or an acoustic output.

According to a further aspect of the invention, the features of the specific feature vector and the predetermined feature vector are a combination of individually determined features of the recognized hand or of the predetermined hand. This can increase the quality of the result of the comparison since more information is taken into account.

According to a further aspect of the invention, the features of the specific feature vector and the predetermined feature vector are determined using an algorithm, preferably using a Gaussian mixed distribution, a combination of classic image features such as color, size, edges, etc., or a neural Network, determined or are been determined. In this way, the corresponding properties and advantages can be used to implement the present invention.

According to a further aspect of the invention, the comparison takes place using a similarity metric between the specific feature vector and the predetermined feature vector, with a match being present if a predetermined sufficient similarity has been determined. In this way, the corresponding properties and advantages can be used to implement the present invention.

According to a further aspect of the invention, the specific feature vector of the identified hand is compared with one or more predetermined feature vectors of a plurality of predetermined hands. This may enable the above-described features and advantages of the present invention to be implemented for a variety of different user types.

According to a further aspect of the invention, the comparison is carried out using a similarity metric between the specific feature vector and several, preferably all, predetermined feature vectors, there being a match with the predetermined feature vector with which the specific feature vector has the greatest similarity having. In this way, the corresponding properties and advantages can be used to implement the present invention.

According to a further aspect of the invention, there is a match with the predetermined feature vector if there is also a predetermined sufficient similarity. A predetermined sufficient similarity can be specified via a corresponding limit value or the like. In this way, it can additionally be ensured that no incorrect user is assigned to the hand currently optically detected just because the match is the least bad for this user. This also allows an unknown user to be recognized. In any case, this can avoid incorrect implementation of the method according to the invention.

• • wenigstens einmaliges, vorzugsweise mehrmaliges, optisches Erfassen wenigstens einer Hand, vorzugsweise beider Hände, eines Benutzers als Bilddaten,
• • Erkennen einer Hand oder zweier Hände eines Benutzers in den optisch erfassten Bilddaten,
• • Bestimmen der Feature-Vektoren der erkannten Hand / Hände,
  • ◯ Zuordnen einer eindeutigen Identifikation zu dem bestimmten Feature-Vektor und bzw. oder
  • ◯ Zuordnen einer Anpassung wenigstens einer Funktion, vorzugsweise einer Parametrierung wenigstens einer Funktion, des Haushaltsgeräts und
• • Speichern der bestimmten Feature-Vektoren als vorbestimmter Feature-Vektor mit der eindeutigen Identifikation und/oder mit der angepassten Funktion, vorzugsweise mit der angepassten Parametrierung der Funktion.

According to a further aspect of the invention, at least one user is pre-determined with the steps:

• • at least once, preferably multiple, optical capture of at least one hand, preferably both hands, of a user as image data,
• • Recognizing a hand or two hands of a user in the optically captured image data,
• • determining the feature vectors of the detected hand/hands,
  • ◯ Assigning a unique identification to the particular feature vector and/or or
  • ◯ Assigning an adjustment of at least one function, preferably a parameterization of at least one function, of the household appliance and
• • Saving the determined feature vectors as a predetermined feature vector with the unique identification and/or with the adjusted function, preferably with the adjusted parameterization of the function.

Such a predetermination can take place before the first use of the method according to the invention for at least one user in order to be able to start using the method according to the invention with at least one predetermined feature vector of a predetermined hand of a predetermined user. However, such a predetermination can also take place in the further use of the method according to the invention in order to add at least one further user.

In any case, the feature vector of the hand of a current user can be determined and this feature vector can then be assigned a unique identification of the current user and/or an adaptation or parameterization of a function to the wishes or preferences of the current user. which can be deposited or stored as predetermined information.

According to a further aspect of the invention, a predetermination takes place if the comparison of the specific feature vector of the recognized hand with at least one predefined feature vector of a predefined hand does not result in a match. Thus, as described above, the predetermination can be carried out or initialized automatically by the method according to the invention if the optically detected hand of the current user cannot be assigned to a predetermined or stored hand. This can increase the convenience for the current user in using the method according to the invention for themselves.

• • Erfassen einer eindeutigen Identifikation des Benutzers und
• • Speichern des bestimmten Feature-Vektors als vorbestimmter Feature-Vektor mit der eindeutigen Identifikation.

According to a further aspect of the invention, at least one user is predetermined if the comparison of the specific feature vector of the recognized hand with at least one predetermined feature vector of a predetermined hand does not lead to a match, with the steps:

• • Capturing a unique identification of the user and
• • Storing the determined feature vector as a predetermined feature vector with the unique identification.

In this case, if the current user cannot be recognized, the current user can be automatically added to the predetermined or stored user by the method, in that the adaptation or parameterization of a function made by the current user is not only carried out but also saved or stored. stored and waiting for the current user to identify themselves to the household appliance or be identified, for example by an operator input or information on the part of the current user, which is assigned to the current user. In this way, the currently performed adjustment or parameterization can be assigned to the function of unique identification of the current user and the current user together with his adjustment or parameterization and optically detected hand or its feature vector can be added to the predetermined or stored information.

The present invention also relates to a household appliance, preferably a kitchen appliance, which is designed to carry out a method as described above. As a result, a household appliance and in particular a kitchen appliance can be made available in order to implement the method according to the invention.

The present invention also relates to a control unit for use in a household appliance as described above, the control unit being designed to carry out a method as described above. This can represent a concrete possibility for implementation.

The present invention also relates to a computer program product with program code for carrying out a method as described above when the computer program product is executed by a household appliance as described above or by a control unit as described above. A computer program product can thereby be made available in order to make the method according to the invention accessible and executable for the household appliance according to the invention as described above or for the control unit according to the invention as described above.

In other words, the present invention is based on the knowledge that the identification of a user on an electronic device, such as on a household appliance, can be a prerequisite for enabling a user-dependent or user-typical device interaction. Identification can be made possible, for example, by a camera built into the front of the device.

Due to the fixed installation position of the camera in the device and the device in the room and the resulting fixed viewing angle of the camera, the use of typical identification methods such as face recognition may not be possible or may be very inconvenient for the user. This means that known identification methods such as face recognition can often only be reached inconveniently for the user due to the rigid housing of a front camera at different installation heights. The camera also has to look into the room in order to recognize the user, and thus also takes on the function of a surveillance camera.

In contrast, hand recognition can allow the identification of a user with an active gesture, such as presenting the open palm of the hand, which can also be possible in a slanted panel and can thus prevent "continuous surveillance".

The invention presented here solves the problem of user identification on an electronic device or on a household appliance by recognizing self-taught features or a combination of features from classic image processing in an image of a user's hand.

The identification can be done using the following method: A feature vector can be calculated from the image of the hand. These features can be a combination of individually calculated features (e.g. color, edge filter, etc.) or they can be learned by an algorithm (e.g. image hashing, Gaussian mixture distribution or neural networks). A database with sample feature vectors from different users or sample images can be created and persistently stored for identification.

If a hand is recognized in a camera image, the feature vector calculated from this hand can be compared with the stored feature vectors from the database. A similarity metric between two vectors (Euclidean distance, cosine distance, Mahalanobis distance) can serve as a comparison. The most similar examples from the database can be used for possible identification and a procedure can be used to assess whether reliable identification is possible. These methods can, for example, consist of thresholding or a self-learning algorithm exist. The feature vectors created by a self-learning algorithm can also make it very easy to add new users.

• 1) Das manuelle Hinzufügen des Nutzers (Manual-Add): Hierbei kann der Nutzer explizit ein Programm zum Hinzufügen eines Nutzers in die Datenbank starten. Dieses Programm kann den Nutzer auffordern, die Hand in die Kamera zu halten, und nimmt dann Beispielbilder der Hand auf. Diese kann mittels eines Hand-Anlernprogramms geschehen. Unter einem Handanlernprogram kann ein Programm verstanden werden, welches den Nutzer auffordert, die Hand in 1-N Posen in die Kamera zu halten, und 1-N Bilder von diesen erfasst und speichert. Dieses Verfahren kann sowohl für eine als auch für zwei Hände durchgeführt werden. Zusätzlich kann der Nutzer eine Identifikations-ID (z.B. Namen) zur Verfügung stellen.
• 2) Das semi-automatische Hinzufügen des Nutzers (Semi-Automatic-Add): Hierbei kann das Handanlernprogramm durch folgendes Verfahren automatisch gestartet werden: Das Gerät kann eine unbekannte Hand erkennen. Eine Hand kann dann als unbekannt angesehen bzw. erkannt werden, falls die Hand durch den Vergleich mit der Feature-Datenbank keinem bekannten Nutzer zugeordnet werden kann (vgl. oben). Bei unbekannten Händen kann nun das Handanlernprogramm gestartet werden.
• 3) Das vollautomatische Hinzufügen des Nutzers (Automatic-Add): Erkennt das Gerät eine Hand im Kamerabild, kann dieses aber nicht einem gespeicherten Nutzer zuordnen werden, muss nicht wie in Punkt 2 das Handanlernprogramm gestartet werden. Stattdessen können das erfasste Bild oder die extrahierten Featurevektoren dann gespeichert werden, wenn diese einem Nutzer durch seine Gerätebedienung eindeutig zuzuordnen werden können (z.B: Auswahl eines Profils). Die Nutzer-ID kann dann automatisch vergeben werden.

For example, the following options exist for creating the user database:

• 1) Manually adding the user (Manual-Add): Here the user can explicitly start a program to add a user to the database. This program can ask the user to put their hand in the camera and then takes sample pictures of the hand. This can be done using a manual training program. A hand tutoring program can be understood to be a program that prompts the user to hold the hand in 1-N poses in the camera and captures and stores 1-N images of these. This procedure can be done for one or two hands. In addition, the user can provide an identification ID (e.g. name).
• 2) The semi-automatic addition of the user (Semi-Automatic-Add): Here the hand training program can be started automatically using the following procedure: The device can recognize an unknown hand. A hand can then be regarded as unknown or recognized if the hand cannot be assigned to a known user through the comparison with the feature database (see above). In the case of unfamiliar hands, the hand training program can now be started.
• 3) The fully automatic addition of the user (Automatic-Add): If the device recognizes a hand in the camera image, but cannot assign it to a saved user, the hand training program does not have to be started as in point 2. Instead, the captured image or the extracted feature vectors can be saved if they can be clearly assigned to a user through their device operation (e.g. selection of a profile). The user ID can then be assigned automatically.

The identification methods presented here can make it possible for the electronic device to recognize the user. For example, the user can be actively identified with the device by showing the palm of the hand. The device can then, for example, display personalized menus, such as frequently selected programs, or load specific configurations. In addition, functions such as child safety devices or authorization to reorder consumables would also be possible. The procedure described when creating the database (see Manual-Add and Semi-Automatic-Add) can make it easy to customize the function at the customer's site.

In addition, the device could also adapt to the user in the background in the case of passive identification of the user (cf. Automatic-Add) by storing the characteristics and certain user-typical device interactions if a hand is reliably recognized and making suggestions in corresponding situations when this hand is recognized .

For the sake of simplicity, it is assumed below that the images of the hand are processed locally on the device by a control device. However, according to the invention, the images can also be transmitted to a cloud, processed there and the result transmitted back to the device. In the following, reference is also made to similarity metrics in order to recognize a user based on the feature vectors. According to the invention, however, this could also be achieved by retraining self-learning algorithms such as support vector machines on the device.

• 1 eine perspektivische Darstellung eines erfindungsgemäßen Haushaltsgeräts als Küchengerät in Form eines Heißgetränkeautomaten; und
• 2 ein Ablaufdiagramm eines erfindungsgemäßen Verfahrens.

An embodiment of the invention is shown purely schematically in the drawings and is described in more detail below. It shows

• 1 a perspective view of a household appliance according to the invention as a kitchen appliance in the form of a hot drinks machine; and
• 2 a flowchart of a method according to the invention.

The 1 is considered in Cartesian coordinates. A longitudinal direction X, which can also be referred to as depth X or length X, extends. A transverse direction Y, which can also be referred to as width Y, extends perpendicularly to the longitudinal direction X. A vertical direction Z, which can also be referred to as the height Z and corresponds to the direction of gravity, extends perpendicularly both to the longitudinal direction X and to the transverse direction Y. The longitudinal direction X and the transverse direction Y together form the horizontal X, Y, which can also be referred to as the horizontal plane X, Y.

1 shows a perspective view of a household appliance 1 according to the invention as a kitchen appliance 1 in the form of a hot drinks machine 1. The hot drinks machine 1 has a front housing 10 in the depth X facing a user on, which can also be referred to as aperture 10. In the lower area of the screen 10 there is a beverage outlet 11, in which vessels 14 can be arranged by the user and filled with the desired hot beverage from the hot beverage machine 1 in the vertical direction Z from above. Precisely the area where the drinks are dispensed 11 can be optically recorded by an optical image recording unit 13 in the form of an optical camera 13 of the hot drinks machine 1 diagonally from the top right.

• Als vorbereitende Maßnahme für die bestimmungsgemäße Verwendung des erfindungsgemäßen Heißgetränkeautomaten 1 kann mittels des erfindungsgemäßen Verfahrens zunächst ein Vorbestimmen 000 wenigstens eines Benutzers erfolgen. Hierzu kann der Benutzer über ein Bedienschnittstelle (nicht dargestellt) des Heißgetränkeautomaten 1 einen entsprechenden Ablauf initialisieren, sodass ein wenigstens einmaliges, vorzugsweise mehrmaliges, optisches Erfassen 010 wenigstens einer Hand 2, vorzugsweise beider Hände 2, des Benutzers mittels der optischen Kamera 13 als Bilddaten erfolgt. Die optisch erfassten Bilddaten können dann von der optischen Kamera 13 an die Steuerungseinheit 12 übergeben werden, um mittels bekannter Verfahren der Mustererkennung in den optisch erfassten Bilddaten die Hand des Benutzers in einem Schritt 030 zu erkennen.

The optically captured image data is transferred from the optical camera 13 to a control unit 12 of the hot beverage machine 1, which not only operates the hot beverage machine 1 but can also process the optically captured image data for this purpose. As a result, a method according to the invention as in the 2 shown as follows:

• As a preparatory measure for the intended use of the hot drinks machine 1 according to the invention, at least one user can first be predetermined 000 by means of the method according to the invention. For this purpose, the user can initialize a corresponding process via an operating interface (not shown) of the hot drinks machine 1, so that at least one hand 2, preferably both hands 2, of the user is optically recorded 010 as image data at least once, preferably several times, using the optical camera 13 . The optically captured image data can then be transferred from the optical camera 13 to the control unit 12 in order to recognize the user's hand in a step 030 using known methods of pattern recognition in the optically captured image data.

A feature vector of the identified hand 2 can now be determined 050 by means of an algorithm such as, for example, a Gaussian mixed distribution, the combination of features from conventional image processing or a neural network. This can reduce the complexity of the method according to the invention, since in the following method steps only the feature vector of the recognized hand 2 has to be processed further instead of the optically recorded image data of the recognized hand 2.

A clear identification can now be assigned 060 to the specific feature vector, so that the current user can be assigned to his optically detected hand 2 or his specific feature vector via his clear identification. If this feature vector is recognized later in the method according to the invention, this user can be inferred as the current user of the hot beverage machine 1 .

Additionally or alternatively, an adaptation 070 of at least one function, preferably a parameterization of at least one function, of the hot beverage machine 1 can be assigned, so that later in the method according to the invention the correspondingly adapted or parameterized function of the hot beverage machine 1 of this optically detected hand or its specific feature vector and can thus be assigned to a specific known user.

In any case, the specific feature vector is now stored 090 as a predetermined feature vector with the unique identification and/or with the adapted function, preferably with the adapted parameterization of the function, which is carried out on the part of a memory unit (not shown) of the control unit 12 the hot drinks machine 1 can be implemented. Thus, the associations described above can be made between a now predetermined hand or its feature vector and a specific predetermined user or its adaptation or parameterization of a function of the hot beverage machine 1 and stored there.

With regular everyday use of the hot beverage machine 1, the current user's hand 2 can approach the beverage outlet 11 of the hot beverage machine 1, for example by means of a proximity sensor (not shown) or the like constantly monitoring this area. If such an approach is recognized, the optical camera 13 can be activated by the control unit 12 in order not to have to be in continuous operation, which could lead to a comparatively high electrical energy consumption. The optical camera 13 can also be activated by switching on the hot beverage machine 1 or by switching the hot beverage machine 1 from a stand-by mode to normal operation.

An optical detection 100 of the beverage dispenser 11 can now take place as an area adjoining the hot beverage machine 1 , so that image data of the beverage dispenser 11 optically captured by the optical camera 13 can be transmitted to the control unit 12 . The control unit 12 can now recognize 200 the hand 2 of the current user in the optically recorded image data of the optical camera 13 . For the recognized hand of the current user, the control unit 12 can then determine 300 a feature vector of the recognized hand 2 take place. The control unit 12 can now compare 400 the specific feature vector of the detected hand of the current user 2 with at least one predefined feature vector of a predefined hand of a predefined user and preferably with at least one predefined feature vector of at least one predefined hand of several predefined ones user done.

A similarity metric between the determined feature vector and the predetermined feature vector can be used for the comparison 400, wherein there can only be a match for a predetermined user if a predetermined sufficient similarity has been determined. If several predetermined users are known and there are thus several predetermined feature vectors of a plurality of predetermined hands, the comparison 400 can be carried out using a similarity metric between the specific feature vector and several, preferably all, predetermined feature vectors, with a match being made with that predetermined feature vector can be assumed, with which the specific feature vector has the greatest similarity. In order not to erroneously assume a predetermined feature vector that matches the least badly as a match, a match with the predetermined feature vector can only be assumed if there is also a predetermined sufficient similarity. If this is not the case, the current user can be recognized as an unknown user and, if necessary, automatically included in the circle of predetermined users by the control unit 12 or the method according to the invention, as will be described in more detail below.

In any case, if the determined feature vector matches the or a predetermined feature vector, an adaptation 500 or a parameterization of the function of the hot beverage machine 1 can be carried out depending on the predetermined hand. Since the predetermined hand is assigned directly or indirectly to a predetermined user and his desired parameterization of the function of the hot beverage machine 1, the corresponding parameterization of the function of the hot beverage machine 1 can be carried out or applied automatically by the control unit 12 by recognizing the current user as a predetermined user , to adapt the hot drinks machine 1 independently to the current user. The current user is thus relieved of having to carry out a desired parameterization of the function of the hot drinks machine 1 .

If the specific feature vector does not match the predetermined feature vector sufficiently, the current operation or use of the hot beverage machine 1 by an unknown user can be automatically used by the control unit 12 to store the current unknown user as a predetermined user. For this purpose, the adaptation or parameterization of a function of the hot drinks machine 1 can be recognized by the control unit 12, implemented and additionally temporarily stored. If the currently unknown user reveals himself, for example, by selecting a profile of the hot drinks machine 1 that is stored and assigned to him, the control unit 12 can detect 600 this unique identification of the user and save 700 the specific feature vector as a predetermined feature Vector with the unique identification of the adjustment or parameterization made by the current user can be stored for future use.

Reference List

X

longitudinal direction; Depth; length

Y

transverse direction; Width

Z

vertical direction; Height

X, Y

Horizontal; horizontal plane

1

household appliance; kitchen utensil; hot drinks machine

10

front housing; cover

11

beverage dispensing

12

control unit

13

optical imaging unit; optical camera

14

vessels

2

hand of a user

000

Predetermine at least one user

010

optically detecting at least one hand of a user

030

Recognizing a user's hand 2 in the optically captured image data

050

Determine a feature vector of the recognized hand 2

060

associating a unique identifier with the particular feature vector

070

Assigning an adjustment to at least one function of the household appliance 1

090

storing the determined feature vector as a predetermined feature vector with the unique identifier and/or with the fitted function

100

Optical detection of an area adjoining the household appliance as image data

200

Recognizing a user's hand 2 in the optically recorded image data

300

Determine a feature vector of the recognized hand 2

400

Comparing the determined feature vector of the recognized hand 2 with at least one predetermined feature vector of a predetermined hand

500

Adjusting the function of the household appliance 1 depending on the predetermined hand

600

Capturing a unique identification of the user

700

storing the determined feature vector as a predetermined feature vector with the unique identification

Claims (15)

Method for adapting a function of a household appliance (1), preferably a kitchen appliance (1), having at least the following steps: • Optical detection (100) of an area adjoining the household appliance (1) as image data, • Recognizing (200) a hand (2) or two hands of a user in the optically captured image data, • determining (300) a feature vector of the recognized hand (2), • Comparing (400) the determined feature vector of the recognized hand (2) with at least one predetermined feature vector of a predetermined hand and • if the determined feature vector matches the predetermined feature vector, adapting (500) the function of the household appliance (1) as a function of the predetermined hand. procedure after claim 1 , wherein the predetermined hand is assigned at least one parameterization of at least one function of the household appliance (1). procedure after claim 1 , wherein the predetermined hand is assigned a predetermined user by means of a unique identification and wherein the predetermined user is assigned at least one parameterization of at least one function of the household appliance (1). Method according to one of the preceding claims, in which the features of the specific feature vector and of the predetermined feature vector are a combination of individually determined features of the recognized hand (2) or of the predetermined hand. Procedure according to one of Claims 1 until 3 , wherein the features of the specific feature vector and the predetermined feature vector are or have been determined by means of a self-learning algorithm, such as image hashing or a neural network. Method according to one of the preceding claims, wherein the comparison (400) takes place by means of a similarity metric between the determined feature vector and the predetermined feature vector, there being a match if a predetermined sufficient similarity is determined. Method according to one of the preceding claims, wherein the comparison (400) of the specific feature vector of the identified hand (2) with a respective predetermined feature vector of a plurality of predetermined hands takes place. procedure after claim 7 , wherein the comparison (400) takes place by means of a similarity metric between the specific feature vector and several, preferably all, predetermined feature vectors, there being a match with the predetermined feature vector with which the specific feature vector has the greatest similarity . procedure after claim 8 , where there is a match with the predetermined feature vector if there is also a predetermined sufficient similarity. Method according to one of the preceding claims, wherein a predetermination (000) of at least one user takes place with the steps: • optically detecting (010) at least one hand (2), preferably both hands (2), of a user as at least once, preferably several times Image data, • Recognizing (030) one hand (2) or several hands of a user in the optically captured image data, • Determining (050) a feature vector of the recognized hand (2), ◯ Assigning (060) a unique identification to the specific Feature vector and/or ◯ assignment (070) of an adaptation of at least one function, preferably a parameterization at least one function of the household appliance (1) and • saving (090) the specific feature vector and/or the image as a predetermined feature vector/image with the unique identification and/or with the adjusted function, preferably with the adjusted parameterization of the function . procedure after claim 10 , wherein a predetermination (000) takes place if the comparison (400) of the specific feature vector of the recognized hand (2) with at least one predetermined feature vector of a predetermined hand does not result in a match. Procedure according to one of Claims 1 until 9 , wherein at least one user is predetermined (000) if the comparison (400) of the specific feature vector of the recognized hand (2) with at least one predetermined feature vector of a predetermined hand does not lead to a match, with the steps: • detecting (600) a unique identification of the user and • storing (700) the specific feature vector as a predetermined feature vector with the unique identification. Household appliance (1), preferably kitchen appliance (1), which is designed to carry out a method according to one of the preceding claims. Control unit (12) for use in a household appliance (1). Claim 13 , wherein the control unit (12) for carrying out a method according to a Claims 1 until 12 is trained. Computer program product with program code for carrying out a method according to a Claims 1 until 12 , if the computer program product after a household appliance (1). Claim 13 or from a control unit (12). Claim 14 is performed.

Images