(19)
(11)EP 3 764 305 A1

(12)EUROPEAN PATENT APPLICATION

(43)Date of publication:
13.01.2021 Bulletin 2021/02

(21)Application number: 19184946.2

(22)Date of filing:  08.07.2019
(51)International Patent Classification (IPC): 
G06Q 30/06(2012.01)
G06K 9/00(2006.01)
G01B 11/08(2006.01)
G01B 11/02(2006.01)
(84)Designated Contracting States:
AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR
Designated Extension States:
BA ME
Designated Validation States:
KH MA MD TN

(71)Applicant: Richemont International SA
1752 Villars-sur-Glane (CH)

(72)Inventors:
  • Grosgogeat, Mathilde
    1205 Genève (CH)
  • Herrera, Bruno
    01210 Ornex (FR)
  • Daguet, Léo
    01170 Gex (FR)

(74)Representative: Poindron, Cyrille et al
Novagraaf International SA Chemin de l'Echo 3
1213 Onex
1213 Onex (CH)

  


(54)CLOSED LOOP OBJECT FITTING TO USER MEMBER


(57) The present invention provides a computer implemented method with a training phase to train a learning machine, and comprises a predicting phase for predicting a predicted fitting opening dimension of a closed loop object to be worn by a user. The training phase comprises the step of supplying to the learning machine pictures of individuals and opening dimensions of a closed loop object worn by the individuals. The predicting phase comprises the step a supplying a picture of a member of a user, and is followed by a step of manufacturing or adjusting or selecting a closed loop object having an opening sized according to the predicted fitting opening dimension.




Description


[0001] The present invention relates to the manufacturing, adjusting, and delivery of closed loop objects, i.e. pieces of jewellery which are intended to be worn by a user around a member of the user. Typically, such closed loop object or piece of jewellery is a ring, or a bangle and thus presents an opening to be positioned around a wearing portion of the member.

[0002] In the present application, a closed loop object to be worn by a user designates an object having an opening for receiving a user's member. Such closed loop object might be rigid, typically made of a non deformable material, such as metal. Thus, the opening of the closed loop object presents a size which might be not adjustable at the user's end. The closed loop object might be a ring or a bangle. However, the closed loop object might present an open loop, but in that case, the open loop is for aesthetic purpose and does not provide any capacity of adjustment to the size of the user's member.

[0003] Document EP3261044A1 discloses the use of body scanner to generate contour map of a user so as to configure products for the user after the body scan. However, this method does not give any hint how to adequately design and manufacture a closed loop object to wear on a wearing portion of a user member adjacent to a joint of same member.

[0004] The present invention aims to address the above mentioned drawbacks of the prior art, and to propose first a method for manufacturing, or for adjusting or for determining the size of a closed loop object to be worn by a user around a user's member which allows a precise fitting so that the closed loop object will be easy to pass over any joint between the wearing portion of the member extremity, but which will also fit closely and comfortably to the wearing portion without any risk of unexpected loss. An object of the above method is also to minimize the manufacturing or rework operations, which typically increase costs or delivery time.

[0005] In this aim, a first aspect of the invention is a computer implemented training method for manufacturing, or for determining the size of, a closed loop object to be worn by a user around a user's member,
wherein the said member comprises:
  • at least one wearing portion where the closed loop object is intended to be worn,
  • a member extremity,
  • at least one joint, arranged between the wearing portion and the member extremity,
wherein the closed loop object comprises at least one opening for passing over the joint,
wherein the training method comprises:
  • a training phase, with the steps of:
    • taking pictures of said member of a plurality of models or individuals, said pictures showing at least the wearing portion and the joint of each model or individual,
    • measuring at least one opening dimension of the opening of a closed loop object worn by each model or individual,
    • supply to a learning machine: the pictures, and the measured opening dimensions, so that the learning machine can:
      • measure a size of the joint and a size of the wearing portion of each model or individual,
      • associate the opening dimension to at least one of a size of the joint and or a size of the wearing portion.


[0006] According to the above embodiment, the computer implemented method comprises a training phase with supplying to a learning machine a series of data issued form real tests related to the member intended to wear the closed loop object, and to the closed loop object, so as to train the learning machine. It has to be noted that the learning machine is supplied with information related not only to the joint, but also to information related to the wearing portion, so that the efficiency and accuracy are increased.

[0007] In practice, a group of persons (i.e. individuals or models) is selected/chosen, and each model or individual has to wear a closed loop object of a given size and a picture showing the member (showing at least the wearing portion and a joint located between the wearing portion and the member's extremity) is taken.

[0008] In particular, the learning machine is supplied with the images of members of individuals to determine or measure the size of the joint and/or of the wearing portion. Then the learning machine can associate the dimension of the opening of the closed loop object worn to the size of the joint and/or of the wearing portion.

[0009] According to an embodiment, the learning machine comprises a neural network or a deep neural network. This kind of learning machine is well suited to image analysis.

[0010] According to an embodiment, the training phase might comprise:
  • recording, for each individual, at least one information related to the fitting of the closed loop object to the model's wearing portion,
so that the learning machine can associate the fitting information to the opening dimension and to the size of the joint and to the size of the wearing portion.

[0011] Each individual gives an information about the fitting of the closed loop to the wearing portion, and the learning machine can also associate the dimension of the opening of the closed loop object to the information about the fitting of the closed loop object to the wearing portion.

[0012] According to an embodiment, the fitting information which might be received from the individual might be chosen from a list comprising at least one of the following criteria:
  • the closed loop object fits adequately the wearing portion,
  • the closed loop object fits too loosely the wearing portion,
  • the closed loop object fits too tightly the wearing portion,
  • the closed loop object does not pass over the joint,
  • the closed loop object is not retained by the joint.


[0013] According to the above embodiment, the learning machine can associate the dimension of the opening to different kinds of fitting information, to increase the possibilities of future prediction, or to lower the amount of persons or individuals required in the group of tester. Instead of having for example 100 individuals each testing one closed loop object, 50 individuals can each test two objects of different sizes if they give the information fit / not fit for each of the two tested closed loop objects.

[0014] According to an embodiment, the step of measuring at least one opening dimension of the opening of a closed loop object worn by each model might comprise the steps of:
  • taking at least one picture of the opening of the closed loop object,
  • supplying the picture to the learning machine so that the learning machine can measure the opening dimension.


[0015] Alternatively or in addition, the step of measuring at least one opening dimension of the opening of a closed loop object worn by each model might comprise the steps of:
  • measuring the opening dimension with a calliper or a ruler, or testing the individual's member with a sizer or gauge to get the dimension of the opening of the closed loop object,
  • supplying the opening dimension to the learning machine.


[0016] The above mentioned sizer or gauge is typically a rigid sheet bearing several openings having different sizes through which a user can introduce its member to determine the adequate size of its member.

[0017] According to an embodiment, each step of taking a picture might include a step of positioning an object or a pattern with a predetermined dimension in the field of imaging.

[0018] According to the above embodiment, the object or pattern positioned in the field of imaging will be recognized by the learning machine and this will allow to accurately determine the scale and size of the other objects shown on the picture.

[0019] According to an embodiment:

the closed loop object might be a ring,

the member might be a finger and preferably the ring finger,

the wearing portion might be a phalanx, preferably the proximal phalanx,

and the joint might be the joint between two phalanges, and preferably the joint between the proximal phalanx and the intermediate phalanx.



[0020] The above method is well suited to determine the size of a ring as the applicant determined that the fitting of the ring to the wearing portion depends on the size of the phalanx joints for allowing the ring to pass over the joints, but the fitting also closely depends on the wearing portion characteristics. In particular, the proximal phalanx shape is an important feature to take into account (width, shape, lean or fat, skin texture....) and a learning machine can take all these features into account. In particular, analysis of the pictures by the learning machine to take all the above mentioned parameters into account is an interesting aspect of the invention.

[0021] According to an embodiment, the learning machine might associate the opening dimension to at least one of:
  • the width of the wearing portion,
  • the width of the joint,
  • the ratio between the width of the wearing portion and the width of the joint,
  • the shape of the member.


[0022] According to an embodiment, the step of taking a picture of the member comprises a step of retro illuminating the member, so as to picture an outside contour of the member, and:
  • a contour or shape of bones of said member, and/or
  • a thickness of skin of said member, and/or
  • a contour or shape of flesh of said member.


[0023] A back light illuminating the member, in the case of a finger, can increase the amount of features detected by the learning machine. In particular, with back lighting, the picture of the user's member can show details of the internal member space, such as bones, muscle, fat, skin thickness. The learning machine can then deduct dimensions or shape, or body characteristics from the picture, to increase the number of parameters taken into account, so that the level of accuracy of future prediction is enhanced.

[0024] According to an embodiment, the learning machine might associate the opening dimension to at least one of:
  • a bone width,
  • a skin thickness,
  • a flesh width.


[0025] A second aspect of the invention relates to a computer implemented predicting method for manufacturing, or for determining the size of, a closed loop object to be worn by a user around a user's member,
wherein the method comprises:
  • a predicting phase, with the steps of:
    • taking at least one picture of a member of a user, said at least one picture showing at least the wearing portion and the joint of the user,
    • supply the picture to a learning machine trained according to the first aspect of the invention, so that the learning machine can:
      • measure a size of the joint and a size of the wearing portion of the user,
      • predict a fitting opening dimension of a closed loop object chosen by the user.


[0026] The predicting phase, occurring after the learning phase, can accurately predict a size of a closed loop object to be worn by the user, without any test with a mock up or test object. In other words, with such prediction, a future manufacturing or adjusting of the size or delivering of the closed loop object can be done directly to end with a comfortable and fitting wearing, with no preliminary physical test for the user. Only one picturing of the user's member is necessary to initiate the predicting phase to end with the predicted fitting opening dimension, which is low burden for the user. The overall process is simplified and new possibilities are offered to the users: it is now possible to select a closed loop object from a catalogue, to take a picture of the member intended for wearing the closed loop object, and to place an order, to receive later the fully finished closed loop object, which will fit with no further adjustment.

[0027] A third aspect of the invention relates to a computer implemented method for manufacturing, or for adjusting the size of, a closed loop object to be worn by a user around a user's member,
wherein the method comprises:
  • manufacturing, or adjusting the size of, a closed loop object according to the fitting opening dimension predicted according to the computer implemented method predicting method of the second aspect of invention, or
  • choosing within a set of pre manufactured closed loop objects a closed loop object according to the predicted fitting opening dimension predicted according to the predicting method of the second aspect of invention.


[0028] Advantageously, the step of taking a picture of the user's member during the predicting phase of second aspect of the invention might be performed by the user, and the picture is sent by the user to a remote server.

[0029] Advantageously, the manufactured or adjusted or chosen closed loop object might be shipped to the user.

[0030] It is to be understood that the above mentioned embodiments might be combined together or dissociated from each other as long as there is no technical inconsistency.

[0031] Other features and advantages of the present invention will appear more clearly from the following detailed description of particular non-limitative examples of the invention, illustrated by the appended drawings where:
  • figure 1 represents a hand of a user wearing two closed loo objects: a ring and a bangle ;
  • figure 2 represents the ring of the figure 1 ;
  • figure 3 represents a hand with typical dimensions to measure and/or to take into account to determine an adequate opening dimension of the ring of figure 2 ;
  • figure 4 represents the hand of figure 3 with other dimensions to measure and/or to take into account to determine an adequate opening dimension of the ring of figure 2 ;
  • figure 5 represents a learning phase of a learning machine ;
  • figure 6 represents a predicting phase of the learning machine.


[0032] Figure 1 represents a hand of a user wearing two closed loop objects CL1 and CL2. The fist closed loop object CL1 is a ring, and the second closed loop object CL2 is a bangle.

[0033] The closed loop object CL1 is positioned on a first wearing portion WP1, which is close to a first joint J1. In the present case, the first wearing portion WP1 is the proximal phalanx of the ring finger, and the first joint is the joint between the proximal phalanx and the medium phalanx.

[0034] The second closed loop object CL2 is positioned on a second wearing portion WP2, close to a second joint J2. In the present case, the second wearing portion WP2 is located in vicinity of the wrist and the second joint J2 is the palm of the hand. However, the second wearing portion could be the distal portion of the forearm, and the second joint could be the wrist or palm or hand of the user.

[0035] Figure 2 shows the closed loop object CL1 (ring) of figure 1. The ring presents an opening having an opening dimension OD, which is in the present case the internal diameter. In the rest of the present application, this opening dimension will be assumed to be the one to be determined accurately for ensuring an adequate fitting to the user member (the ring finger of figure 1). However, the present method can also consider other dimensions to be relevant to ensure an adequate fitting, such as the width of the ring, for instance.

[0036] Adequate fitting in the present case means that the ring is sized so as to at least:
  • allow easy positioning of the ring onto the first wearing portion WP1, i.e. to pass easily with no difficulty over the first joint J1,
  • stay stably over the first wearing portion WP1, i.e. the ring will not fall unexpectedly off the finger,
  • not generate pain or discomfort to the user because it is too tightly adjusted onto the first wearing portion WP1. The same holds for the second closed loop object CL2: the opening dimension shall be sized adequately to provide a comfortable and secured wearing.


[0037] In that aim, the method of the present invention proposes to take into account some dimensions of the member (hand in the present case) so a to predict a predicted fitting opening dimension.

[0038] In particular, and as shown on figure 3, the method can take into account several dimensions of the user's member, such as the width WP1D of the first wearing portion WP1, the width J1D of the first joint J1... for determining the adequate size of the first closed loop object CL1. For the second closed loop object CL2, the width WP2D of the second wearing portion WP2 and/or the width J2D of the second joint J2 might be considered.

[0039] Pictures of the hand can be easily taken and then analysed by a software (a neural network or a deep neural network) to obtain the relevant dimensions. As an example, feedforward neural network might be used, and in particular a multilayer perceptron can be used to implement the training method according to the present invention.

[0040] Other methods are also encompassed to get the dimensions of the joint and wearing portion, such as range imaging (stereo triangulation with stereo or multiple camera setup, time of flight technique, structured light...). Devices such as body scanner, Lidar, laser range finders, stereo or multiple camera can be used.

[0041] The picturing might be done by the user, or by a vendor and might be sent to a remote server/computer for analysis.

[0042] It is also encompassed by the present method to take other information/dimensions into account. As shown on figure 4, the internal structure of the hand might be analysed, with, for example, a back light illuminating the hand during the picturing. The picture will then show internal bones or shapes (see dotted lines on figure 4) for the computer analysis so that the machine learning can measure and take into account bones width/dimensions and deduct from the external dimensions a ratio of muscle or fat to increase the accuracy of future prediction.

[0043] In particular, the method can take into account the bone dimension B1D of the proximal phalanx and the bone dimension B2D at the joint between the proximal phalanx and the middle or medium phalanx. From these measures, it is possible to consider various ratios between the bone dimensions and outer dimensions of the finger such as above mentioned dimensions J1D and/or WP1D to deduct a muscle ratio, fat ratio, or even skin thickness.

[0044] All these parameters or dimensions or ratios can be taken into account to predict a fitting opening dimension for the closed loop object to be worn onto the wearing portion.

[0045] Figure 5 shows how a learning machine LM can be trained to give predictions in a next phase. The training method can be based on selecting a group of individuals for implementing the following steps for each individual of the group. A picture or a series of pictures can be taken to show the hand of each individual and in particular, in case the closed loop object is a ring, to show the first wearing portion WP1 (the proximal phalanx) and the first joint J1.

[0046] Each individual will also wear a closed loop object (a ring) which is previously measured and which is chosen to fit the wearing portion in comfortable way. In case of a ring, the internal diameter can be measured with a calliper, with a ruler or with a sizer gauge (a plate with multiple holes having given sizes). Pictures of the closed loop object CL1, CL2 can also be used.

[0047] All the pictures for each individual and the opening dimension measured are submitted to a learning machine LM. In detail, for the first individual, picture P1 and opening dimension OD1 are submitted. For the second individual, picture P2 and opening dimension OD2 are submitted, and the same for each individual of the group.

[0048] The pictures can be taken with a reference object or pattern with known size, to allow easy recognition and scaling for the learning machine LM.

[0049] With these data, the learning machine LM can extract from the pictures all the relevant dimensions, shapes, ratios above mentioned, and associate this picture information to the corresponding opening dimension provided to the learning machine LM. For example, a table T can be generated.

[0050] In addition, each individual can also give fitting information (if the ring is loose, or tight or lost during experience...). This fitting information can be also supplied to the learning machine to be included in the table T.

[0051] Figure 6 illustrates a predicting phase implemented once the learning machine LM has been trained with a sufficiently large group of individuals (50 persons, or 100 persons or more).

[0052] During this predicting phase, a user hand is pictured, and this picture P is sent to the learning machine LM. Analysis of the picture P is performed to extract the relevant data (above mentioned dimensions such as WP1D, J1D, B1D, B2D..., shape of the hand, ratios, ...) and the learning machine LM can output a predicted fitting opening dimension PFOD for a ring that will adequately fit to the user hand, with no physical test.

[0053] Since the prediction is done by taking into account one or several of the above mentioned parameters, the predicted fitting opening dimension is more accurate and customized to the user morphology.

[0054] During the predicting phase, pictures of the hand can be easily taken and then analysed by a software (a neural network or a deep neural network) to obtain the relevant dimensions. As an example, feedforward neural network might be used, and in particular a multilayer perceptron can be used to implement the predicting method according to the present invention.

[0055] Other methods are also encompassed to get the dimensions of the joint and wearing portion, such as range imaging (stereo triangulation with stereo or multiple camera setup, time of flight technique, structured light...). Devices such as body scanner, Lidar, laser range finders, stereo or multiple camera can be used.

[0056] Finally, the prediction of the predicted fitting opening dimension might be done by a software (a neural network or a deep neural network). As an example, feedforward neural network might be used, and in particular a multilayer perceptron can be used.

[0057] Consequently, manufacturing of a ring can be done for further delivery to the user, after the order is placed. Other possibilities are a step of adjusting the size of the ring, or selecting a ring within a pre-manufactured set of rings.

[0058] At this stage, it has to be understood that the training phase and the predicting phase allow to give a predicted fitting opening dimension PFOD to manufacture or adjust or select the adequate ring. In other words, it is not needed anymore to manufacture in advance sets of rings for testing or adjusting a given ring for a customer. The manufacturing or adjusting or selecting step is done after the user hand has been pictured and analysed to give in one go the adequate predicted fitting opening dimension.

[0059] Such process enables to reduce the stock required in retailing shops, and/or increases the accuracy of manufacturing - adjusting for perfect fitting and/or enables to place an order via internet for receiving a closed loop object (a ring) which fits the user member (finger).

[0060] It is of course understood that obvious improvements and/or modifications for the one skilled in the art may be implemented, still being under the scope of the invention as it is defined by the appended claims.

[0061] In particular, only the internal diameter of the opening of the closed loop object has been described in detail, but the method can also deal with other dimensions such as the thickness of the closed loop object, connecting radiuses between surfaces. Other parameters can also be taken into account, such as the weight, the material, the roughness.

[0062] It has also to be mentioned that the method can also be used for non circular objects.

[0063] Other members can be also analyzed, such as the leg and foot, where closed loop objects can also be worn.

[0064] The above specification mentions that pictures are taken, but it has to be taken in mind that the present invention can also deal with movies, which are a series of pictures. Black and white pictures/movies can be taken, but color images can also be used.


Claims

1. Computer implemented training method for manufacturing, or for determining the size of, a closed loop object (CL1, CL2) to be worn by a user around a user's member,
wherein the said member comprises:

- at least one wearing portion (WP1, WP2) where the closed loop object (CL1, CL2) is intended to be worn,

- a member extremity,

- at least one joint (J1, J2), arranged between the wearing portion (WP1, WP2) and the member extremity,

wherein the closed loop object (CL1, CL2) comprises at least one opening for passing over the joint (J1, J2),
wherein the training method comprises:

• a training phase, with the steps of:

- taking pictures of said member of a plurality of individuals, said pictures showing at least the wearing portion (WP1, WP2) and the joint (J1, J2) of each individual,

- measuring at least one opening dimension of the opening of a closed loop object (CL1, CL2) worn by each individual,

- supply to a learning machine: the pictures, and the measured opening dimensions, so that the learning machine can:

- measure a size of the joint (J1, J2) and a size of the wearing portion (WP1, WP2) of each individual,

- associate the opening dimension to at least one of a size of the joint (J1, J2) and or a size of the wearing portion (WP1, WP2).


 
2. Computer implemented training method according to claim 1, wherein the training phase comprises:

- recording, for each individual, at least one information related to the fitting of the closed loop object (CL1, CL2) to the individual's wearing portion (WP1, WP2),

so that the learning machine can associate the fitting information to the opening dimension and to the size of the joint (J1, J2) and to the size of the wearing portion (WP1, WP2).
 
3. Computer implemented training method according to claim 2, wherein the fitting information is received from the individual and is chosen from a list comprising at least one of the following criteria:

- the closed loop object (CL1, CL2) fits adequately the wearing portion (WP1, WP2),

- the closed loop object (CL1, CL2) fits too loosely the wearing portion (WP1, WP2),

- the closed loop object (CL1, CL2) fits too tightly the wearing portion (WP1, WP2),

- the closed loop object (CL1, CL2) does not pass over the joint (J1, J2),

- the closed loop object (CL1, CL2) is not retained by the joint (J1, J2).


 
4. Computer implemented training method according to anyone of claims 1 to 3, wherein the step of measuring at least one opening dimension of the opening of a closed loop object (CL1, CL2) worn by each individual comprises the steps of:

- taking at least one picture of the opening of the closed loop object (CL1, CL2),

- supplying the picture to the learning machine so that the learning machine can measure the opening dimension.


 
5. Computer implemented training method according to anyone of claims 1 to 4, wherein each step of taking a picture includes a step of positioning an object or a pattern with a predetermined dimension in the field of imaging.
 
6. Computer implemented training method according to anyone of claims 1 to 5,
wherein the closed loop object (CL1, CL2) is a ring,
wherein the member is a finger and preferably the ring finger,
wherein the wearing portion (WP1, WP2) is a phalanx, preferably the proximal phalanx,
and wherein the joint (J1, J2) is the joint between two phalanges, and preferably the joint between the proximal phalanx and the intermediate phalanx.
 
7. Computer implemented training method according to anyone of claims 1 to 6, wherein the learning machine associates the opening dimension to at least one of:

- the width (WP1D, WP2D) of the wearing portion (WP1, WP2),

- the width (J1D, J2D) of the joint (J1, J2),

- the ratio between the width of the wearing portion (WP1, WP2) and the width of the joint (J1, J2),

- the shape of the member.


 
8. Computer implemented training method according to anyone of claims 1 to 7, wherein the step of taking a picture of the member comprises a step of retro illuminating the member, so as to picture an outside contour of the member, and:

- a contour or shape of bones of said member, and/or

- a thickness of skin of said member, and/or

- a contour or shape of flesh of said member.


 
9. Computer implemented training method according to claim 8, wherein the learning machine associates the opening dimension to at least one of:

- a bone width,

- a skin thickness,

- a flesh width.


 
10. Computer implemented predicting method for manufacturing, or for determining the size of, a closed loop object (CL1, CL2) to be worn by a user around a user's member,
wherein the method comprises:

• a predicting phase, with the steps of:

- taking at least one picture of a member of a user, said at least one picture showing at least the wearing portion (WP1, WP2) and the joint (J1, J2) of the user,

- supply the picture to a learning machine trained according to the computer implemented training method of anyone of claims 1 to 9, so that the learning machine can:

- measure a size of the joint (J1, J2) and a size of the wearing portion (WP1, WP2) of the user,

- predict a fitting opening dimension of a closed loop object (CL1, CL2) chosen by the user.


 
11. Computer implemented method for manufacturing, or for adjusting the size of, a closed loop object (CL1, CL2) to be worn by a user around a user's member,
wherein the method comprises:

- manufacturing, or adjusting the size of, a closed loop object (CL1, CL2) according to the fitting opening dimension predicted according to the computer implemented method predicting method of claim 10, or

- choosing within a set of pre manufactured closed loop objects a closed loop object (CL1, CL2) according to the predicted fitting opening dimension predicted according to the predicting method of claim 10.


 
12. Computer implemented method for manufacturing, or for adjusting the size of, a closed loop object (CL1, CL2) according to claim 11, wherein the step of taking a picture of the user's member during the predicting phase of claim 10 is performed by the user, and the picture is sent by the user to a remote server,
and wherein the manufactured or adjusted or chosen closed loop object (CL1, CL2) is shipped to the user.
 




Drawing













Search report









Search report




Cited references

REFERENCES CITED IN THE DESCRIPTION



This list of references cited by the applicant is for the reader's convenience only. It does not form part of the European patent document. Even though great care has been taken in compiling the references, errors or omissions cannot be excluded and the EPO disclaims all liability in this regard.

Patent documents cited in the description