Method for modeling and assessing the progression of signs of aging in a user - Patent Application 20070122997

Abstract

The present invention provides a computer-implemented method for modeling and assessing the progression of signs of aging (S1, S2, S3, S4) in a user, comprising: - receiving data corresponding to a set of pixels of an image (Pi) of a body area of ​​the user generally exhibiting at least one visible sign of aging; -Severity score (Ns i ) to visible signs of aging of said body region; - at least one age-related factor (f i ) as a function of the value of the severity score (N's i and modifying the

Description

[Technical Field]

[0001] The present invention relates to a method for modeling, predicting and assessing the progression of signs of skin aging in a user, and a corresponding computer-implemented system for virtually simulating said progression on an image of the user. [Background technology]

[0002] Many people are interested in knowing how much their skin is likely to change over time, and in particular in showing more noticeable visible signs of age.Signs of skin aging are understood to mean, for example, wrinkles or brown spots.This knowledge may be simply out of curiosity, or may be aimed at taking preventive measures, especially by applying cosmetic or therapeutic treatments, or even by changing their lifestyle habits whenever possible.

[0003] Indeed, the presence of wrinkles or blemishes on the skin, especially on skin surfaces that are generally not covered and therefore visible, such as the face or hands, often poses an aesthetic problem that may be more or less difficult for different individuals to accept.

[0004] The progression of such signs of aging over time depends on many factors, both internal (skin type, genetics) and external (pollution, UV exposure, smoking) that are specific to the individual, which can be difficult to understand and can complicate such predictions.

[0005] One relatively recent method, still in use, involved the use of an atlas containing one or more sets of photographs of different skin types at different stages in time (Skin Aging Atlas published in five volumes: (Non-Patent Document 1), (Non-Patent Document 2), (Non-Patent Document 3), (Non-Patent Document 4), (Non-Patent Document 5)).

[0006] One of the main purposes of these atlases is to allow an objective assessment of facial aging. These books study the skin according to age and ethnic type, thus systematically establishing its characteristics and classification, allowing the assignment of severity scores (generally ranging from 0 to 9) to different areas of the face (open pores, frown lines, crow's feet, sagging neck, etc.). The classifications can also be used together to assess overall facial aging.

[0007] These atlases are primarily intended for professionals and are primarily intended to allow an objective assessment of the results of "anti-aging" procedures aimed at blending or masking these signs by comparing before / after severity (topical procedures such as, for example, botulinum toxin injections into crow's feet, but also more common procedures, for example, of the lifting or peeling type).

[0008] The development of computer processing tools and the deployment of artificial intelligence methods have made it possible to automate these assessments. For example, in January 2019, VICHY was able to offer a tool called SKINCONSULT AI that allows consumers to obtain a digital diagnosis of their skin based on a portrait photograph.

[0009] This tool is based on an artificial intelligence algorithm developed by MODIFACE and trained using the set of L'OREAL images specifically used to form the aforementioned atlas. For further details, see U.S. Patent No. 5,929,999.

[0010] Therefore, based on a portrait photo, the tool can assess seven signs of aging: wrinkles around the eyes, lack of firmness, fine lines, lack of skin tone, pigmentation spots, deep wrinkles, and pores, and provide users with product recommendations.

[0011] Although easily used by the general public, this tool, like the atlas, provides an assessment at a given moment and does not allow any assessment of how these signs of aging may progress over time.

[0012] Artificial intelligence computing techniques have also been used in many applications, particularly in methods designed to provide users with predictions of potential facial progression over time by altering photographs to simulate the likely appearance at a given age.

[0013] Examples of such methods are described in particular in WO 02 / 04999, ... and in particular in WO 02 / 04999 and WO 02 / 04999, which serve as further useful references.

[0014] The document (Patent Document 2) describes a method of comparing aging data obtained on a user at a first time point and a second time point (e.g., after 6 months) to extrapolate subsequent progression (e.g., after 18 months and 66 months). Such a method is limited to a few months and is difficult to use to simulate progression that may occur over 10, 15, or 20 years.

[0015] A similar approach is used in (Patent Document 3) in that photographs of a person at two different ages are used to virtually transfer the detected progression to the aging of another person. A similar method is described in (Non-Patent Document 6).

[0016] Patent document 4 relates to a method for predicting the appearance of an external part of a human body as a function of time and / or treatment, in which at least three images are generated, which correspond to different scores of at least one aspect parameter as a function of time and / or treatment, and the variation of at least this aspect parameter on said images is non-linear. Preferably, the variation of said aspect parameter occurs according to a progression law determined based on observations of this aspect parameter carried out in a reference population. Therefore, the predicted progression is not actually personalized, but merely estimates aging markers of different severity scores for the user.

[0017] To better personalize the prediction of the progression of signs of aging, the patent application (Patent Document 5) aims to integrate data on the user's lifestyle and habits, in particular to take into account several external factors of aging, such as sun exposure, sleep, smoking, etc. It is stated that machine learning means are implemented to predict the possible progression of aging. However, few details are provided, and the patent application merely lists a set of potentially usable learning models.

[0018] In practice, it is extremely difficult, if not impossible, to conventionally train an artificial intelligence engine to predict the temporal progression of visible signs of aging over a period of several years, and indeed it is difficult to imagine having sufficient training data (and in particular "ground truth" for each parameter) over such a timescale.

[0019] Therefore, it is generally necessary to resort to additional theoretical modeling and use average progression across a given age cohort.

[0020] This is provided in particular in the document (Patent Document 6), which provides for detecting the ethnic type of the subject and applying the corresponding average aging model, but such a method does not take into account the individualization and external factors of individualized aging.

[0021] Thus, the viral application "FaceApp" is known, which uses self-portrait photos to generate an image of what the user is likely to look like when they are older. The application uses a series of transformations applied to different distinctive elements of the face, trying to preserve as much of the individual's features and characteristics as possible. The transformations are obtained by training a cGAN-type artificial intelligence engine based on defined age groups (https: / / analyticsindiamag.com / the-ai-behind-faceapp / & https: / / iq.opengenus.org / face-aging-cgan-keras / ).

[0022] For further details on such techniques, see the articles (Non-Patent Document 7) (https: / / doi.org / 10.48550 / arXiv.1702.01983) and (Non-Patent Document 8).

[0023] The models obtained in this way remain extremely limited in terms of personalization, especially since, when the number of conditions is increased, the amount of available training data decreases significantly to the point where it becomes insufficient, and generally allow the application of statistical transformations that do not take into account parameters related to the user's lifestyle.

[0024] The "ChangeMyFace" application (https: / / changemyface.com / ) proposes to take into account parameters related to the user's lifestyle. Possible parameters include diet, alcohol consumption, physical activity level, stress level, exposure to polluted environments, exposure to sunlight, amount of sleep, and tobacco consumption. However, no information is provided on how these parameters are taken into account to influence the modification of the user's photo.

[0025] The paper by (Non-Patent Document 9) (https: / / tel.archives-ouvertes.fr / tel-03467715) aims to take into account external factors such as sun exposure, tobacco consumption, air pollution, and individual intrinsic factors such as a person's ethnicity and their hormone levels. The aim of this work is to model the influence of these parameters and incorporate them into the rendering image generator. However, this can be particularly complex and increases the number of parameters to be integrated into the model. [Prior art documents] [Patent documents]

[0026] [Patent Document 1] International Publication No. 2020 / 113326A1 Brochure [Patent Document 2] European Patent Application Publication No. 1298562A1 [Patent Document 3] U.S. Patent No. 4,276,570 [Patent Document 4] French Patent Application Publication No. 2875930A1 [Patent Document 5] European Patent Application Publication No. 3699811A1 [Patent Document 6] US Patent No. 2018 / 276869 [Non-patent literature]

[0027] [Non-Patent Document 1] Roland Bazin, Frederic Flament, Huixia Qiu:Skin Aging Atlas.Volume 5,Photo-aging Face & Body.October 2017. [Non-patent document 2] Roland Bazin, Frederic Flament, Virginie Rubert:Skin Aging Atlas.Volume 4, Indian Type.June 2014. [Non-patent document 3] Roland Bazin, Frederic Flament, Franck Giron:Skin Aging Atlas.Volume 3, Afro-American type.May 2012. [Non-patent document 4] Roland Bazin,Frederic Flament:Skin Aging Atlas.Volume 2,Asian type.November 2010. [Non-patent document 5] Roland Bazin:Skin Aging Atlas Volume 1,Caucasian Type.October 2007. [Non-patent document 6] Lanitis A.,et al.,Title 'MODELING THE PROCESS OF AGEING IN FACE IMAGES',Computer Vision,1999,The Proceedings of the Seventh IEEE International Conference on Kerkyra,Greece,20-27 Sept.1999,Los Alamitos,CA,USA,IEEE Comput.Soc.,US,Vol.1,20 September 1999,pages 131-136. [Non-Patent Document 7] Antipov, G.; Baccouche, M.; Dugelay, J.-L., Title 'FACE AGING WITH CONDITIONAL GENERATIVE ADVERSARIAL NETWORKS', Proceedings of the 2017 IEEE International Conference on Image Processing(ICIP), Beijing, China, 17-20 September 2017; pages 2089-2093. [Non-patent document 8] Xinhua Liu, Title 'BIDIRECTIONAL FACE AGING SYNTHESIS BASED ON IMPROVED DEEP CONVOLUTIONAL GENERATIVE ADVERSARIAL NETWORKS', MDPI Information 2019,10,69;doi:10.3390 / info10020069. [Non-Patent Document 9] Farnaz Majid Zadeh Heravi, Title 'THREE-DIMENSION FACIAL DE-AGEING AND AGEING MODELING:EXTRINSIC FACTORS IMPACT', Signal and Image processing, Universite Paris-Est, 2019. Summary of the Invention [Problem to be solved by the invention]

[0028] Therefore, there is a requirement to develop methods and systems to improve both the accuracy of progress predictions and the ability to personalize these progress predictions for the user. The methods must also remain relatively simple so that they can be easily implemented via a portable personal computer terminal. [Means for solving the problem]

[0029] To this end, the present invention provides - receiving data corresponding to a set of pixels of an image of a body area of ​​a user that generally exhibits at least one visible sign of aging; - assigning a severity score to the visible signs of aging of said body area; - modifying the severity score as a function of the value of at least one aging factor; We propose a method including:

[0030] The method is fully or partially computer-implemented.

[0031] The term "generally exhibiting signs of aging" is understood to mean that the target body area is an area where the considered signs of aging appear and appear in the general population, even if the signs are not yet clearly present in the user who receives the method. Thus, even if the user is not yet old enough to have visible crow's feet, the area where crow's feet occur is well known, and the degree of severity is simply zero or one (or the minimum value on the scale). The same applies to the other aforementioned visible signs of aging. Of course, the image of the considered body area can be obtained from an image covering a larger body area that encompasses the body area of ​​interest, which can then be separated by a segmentation or classification method based on body markers (for example, computer image analysis can be used to separate the body area where crow's feet appear based on the identification of the pupil and the edge of the eye).

[0032] In particular, the visible signs of aging are selected from enlarged skin pores, wrinkles between the eyebrows, crow's feet, sagging neck, wrinkles around the eyes, firmness or sagging of the skin, the presence of fine lines, complexion or skin lightness, the presence of pigmentation spots.

[0033] Severity score or degree of severity is understood to mean a numerical value forming part of an ordinal scale. The scale may include a minimum value, in particular zero or 1. The scale may also include a maximum value. As mentioned above, conventional atlases set a maximum value of 9. The progression of degrees on the severity scale can be expressed as integers (0, 1, 2, 3, ..., 9) or decimal numbers, in particular in increments of 0.5 degrees (0, 0.5, 1, 1.5, ...).

[0034] Therefore, by assigning a severity score to each aging sign, this score can be used to perform simple calculations, particularly additive calculations, by increasing or decreasing the severity score initially assigned as a function of the value assigned to one or more aging factors according to their impact on the aging sign, particularly according to the user's lifestyle habits.Then, the modified severity score corresponds to the prediction of the progression of the considered aging sign over time.For example, for a user whose skin is rated as having a wrinkle severity level of 5, it can be possible to estimate the severity level after 10, 15 or 20 years by simply adding or subtracting the level according to the values, preferably personalized values, assigned to various considered aging factors.

[0035] The modified severity score may then be used as input data for an image generator configured to transform the initial image as a function of input parameters corresponding to the severity score, thus generating a modified image presenting the considered signs of aging with the desired modified severity.

[0036] Therefore, advantageously in a complementary manner, the method comprises the additional step of generating a modified image of the user's body area having visible signs of ageing corresponding to the modified severity score, which can advantageously be displayed on a screen for presentation to the user.

[0037] The modified severity score may or may not be displayed separately or together with the modified image for presentation to the user.

[0038] The initial severity score and / or initial images may also be displayed for presentation to the user for comparison of any progress. To facilitate comparison, the severity scores (initial and corrected) and / or images (initial and corrected) may be displayed separately or side-by-side.

[0039] Preferably, the method comprises the use of at least one user-independent (or not user-specific) aging factor, such as the time, in particular the duration, over which the progression of the considered signs of aging is intended to be estimated. This period is preferably more than 1 year, preferably more than 6 years, or even more than 10 years. Preferably, a 10-, 15-, or 20-year prediction will be sought.

[0040] Thus, for example, a period of 15 years can be considered as increasing the wrinkle severity by 2 degrees (+2), or in the same example, the wrinkle severity of the skin reaches a value of 7.

[0041] Preferably, the method includes the use of at least one user-dependent (or user-specific) ageing factor. In that case, the method advantageously includes the additional step of receiving data corresponding to user-adapted personalized values ​​for all or some of said user-dependent ageing factors. In the absence of data for a particular factor, default values ​​can be provided.

[0042] Advantageously, the personalized value may correspond to the user's current state.

[0043] Complementarily or optionally, the method can be implemented using personalized data corresponding to a changed state of the user (counterfactual approach, e.g., if a smoking user quit smoking, or if a user who does not use moisturizing cosmetics adopted such products).

[0044] The various modified severity scores and, if applicable, the corresponding modified images, according to the changes made in the values ​​of the aging factors taken into account, may be displayed and presented to the user for comparison.

[0045] According to a first embodiment, the various ageing factors have a limited number of possible values, preferably a maximum of 2 or 3 or even 4. The proposed values ​​advantageously reflect the strength of the presence of the considered ageing factor in the user.

[0046] According to a second embodiment, all or some of the various aging factors each have possible values ​​associated with a weighting factor specific to the user. Thus, for example, the user may provide extreme values ​​and be asked to associate weighting factors with them, allowing them to represent specific intermediate intensities. These weighting factors are reported and used to weight the influences and variations determined for each extreme value to obtain an intermediate influence adapted to the user.

[0047] According to certain embodiments, user-independent factors, in particular time, are set and are not personalized for the user. Alternatively, their values ​​can be modified, in particular by the user or operator. However, it is preferable to use a set duration that cannot be changed, especially for the time factor. In fact, the influence established for user-specific aging factors (each factor or pooled across several factors) can also take the time factor into account, and implementing a modifiable time factor requires a lot of resources to adapt the influence of the user-specific factors not only as a function of the values ​​given by the user for these factors, but also as a function of the intended duration of the estimation.

[0048] According to one embodiment, the severity score is modified by applying a computer model configured to determine at least one variation of the severity score as a function of the values ​​assigned to the aging factors, said variation being added to the initial severity score assigned to the considered sign of aging.

[0049] Preferably, the method includes an additional step of receiving information corresponding to the user's skin type and applying a corresponding adapted computer model. In particular, models of the variation or progression of signs of aging can be provided for each ethnicity type, i.e., a Caucasian skin progression model, an African skin progression model, and an Asian skin progression model, among others. The user's skin type can be provided manually or can be established from an image analysis step aimed at determining, for example, skin color parameters, as can be achieved in particular in applications for determining foundations. This step is preferably performed on received image data of the body area in which the progression of signs of aging is to be predicted. Similarly, data related to the user's gender and / or their age (or membership in an age group) can also be provided to take into account; these data can be entered manually or can be the result of an image analysis and evaluation process for applying a model corresponding to the criteria taken into account.

[0050] According to an advantageous embodiment, it is possible to apply several models specifically designed for different types of users and weight the results. Such a method is particularly useful for the user's age. In fact, progression models are generally designed for different age ranges of users (it would be difficult to design one model for each age). However, strictly applying the model for the age range to which the user belongs would be too simplistic and would carry the risk of underestimating or overestimating the progression, especially depending on whether the user's age is relatively close to the lower or upper limit. Therefore, according to an alternative embodiment, the method includes an additional step of obtaining one or more weighting factors, for example via an interpolation function, to assign to the results of one or more predictive models applied according to the user's type. By applying each weighting factor to the results of each model, it is possible to obtain the final variation to be applied to the initial severity.

[0051] Advantageously, this step is carried out on the age of the user, and the step aims to obtain, as a function of the age of the user, weighting coefficients to be applied to the different predictive models established by age categories. The predictive models (all or at least those that return a weighting coefficient that is not zero or significant) are applied to the initial severity and the results obtained are weighted with coefficients determined by the interpolation function to obtain a modified severity.

[0052] According to certain embodiments, the computer model is a probabilistic model that returns at least one variation of the severity score and an associated probability. Preferably, the computer model returns several variations of the severity score, each having an associated probability. Thus, the method can determine several modified severity scores, each modified severity score corresponding to the initial severity score to which the variation was added. Each modified severity score is associated with a probability corresponding to the added variation. Modified images corresponding to each modified severity score can be generated for display purposes for presentation to the user. Preferably, only modified images corresponding to the variation associated with the highest probability are generated. Alternatively, an average modified severity score can be determined by weighting the variations by their associated probabilities.

[0053] Advantageously, the probabilistic computer model is a Bayesian network, in particular a causal Bayesian network. Advantageously, the Bayesian network can be constructed from the knowledge of experts implementing elicitation methods and / or causal inference techniques.

[0054] The variation in severity determined by the model may result from calculations performed on the determined effects of one or more aging factors depending on the values ​​assigned to the one or more aging factors. In particular, an initial effect can be identified for each aging factor considered, and then, advantageously, the effects can be pooled by grouping several aging factors. Advantageously, aging factors are grouped into different categories, such as beneficial factors, degradative factors, intrinsic factors, extrinsic factors, and especially environmental factors, in particular to take into account any synergistic effects between the above factors, if applicable. The effects of the various aging factors depending on their values ​​can be identified empirically and / or theoretically. Depending on the model used, the effects can be single-factor or multi-factorial.

[0055] Thus, if the user is a habitual smoker, this progression can be exacerbated by further increasing the severity of skin wrinkles by one or two degrees. A user who is a habitual smoker can be determined to be at risk after 15 years of having skin with a wrinkle severity of 8 (initial severity of 5, +2 for the time factor set at 15 years, +1 for the "smoker" factor with a "yes" value).

[0056] Based on the estimated degree of severity, it is advantageous to modify the initial image accordingly in order to render to the user an estimated image of the user's skin in 15 years' time.

[0057] Advantageously, the method is performed on image data of a body region exhibiting several visible signs of aging, with each sign of aging having been assigned a severity score in accordance with the method by applying a computer model specifically tailored to the progression of each sign of aging, and a single modified image of all signs of aging, each with its associated modified severity score, if applicable, is generated and displayed for presentation to a user.

[0058] Advantageously, the assignment of an initial severity to the considered visible signs of aging is carried out by image analysis, with or without a prior segmentation of the region of interest.

[0059] In particular, the severity level is assigned by a trained artificial intelligence engine, such as the aforementioned SKIN CONSULT. Alternatively or additionally (see not yet published French Patent No. 22 / 04419), image analysis can be used to measure characteristic physical parameters from the optical properties of the area of ​​interest. Characteristic physical parameters can be, for example, the number, density, length, and / or depth of wrinkles (e.g., by analyzing contrast in a known manner), the color or contrast of the skin area, and, for example, the color, number, surface, and / or density of skin blemishes. The values ​​of the measured physical parameters can then be used to assign a severity level to the considered sign of aging.

[0060] Advantageously, the method includes a preceding step of acquiring a digital image of a body region, preferably the body region being the face.

[0061] Preferably, the images of the user's body area are acquired under standardized conditions, in particular under standardized lighting conditions, in particular under D65 lighting conditions. Otherwise, in particular in the case of self-portraits or selfies taken by the user themselves using a mobile device or smartphone, an image correction / normalization algorithm may be applied before the initial severity is assigned by the image analysis algorithm.

[0062] Of course, an inverse transform can be applied to the modified image before display and presentation to the user in order to render to the user an estimated image corresponding to the same image capture conditions.

[0063] The invention also relates to a system for implementing the method, in particular a computer system for implementing said method on a computer.

[0064] In particular, the system comprises at least one first data input configured to receive data corresponding to a set of pixels of a digital image, and a second data input configured to receive at least one value of an aging factor.

[0065] Preferably, the system comprises a digital photograph acquisition device in communication with the first data input.

[0066] Preferably, the system also comprises a human-machine interface, in particular a keyboard or touch screen, connected to a second data input and enabling a user or operator to input one or more values ​​for one or more ageing factors.

[0067] The system also includes at least one processor configured to assign a severity score of at least one visible sign of aging from the received pixels.

[0068] According to the present application, the processor is also configured to modify said severity score as a function of the received value of the aging factor and return said modified severity score.

[0069] Advantageously, the system additionally includes an image generator configured to generate, from the received image, a modified image representative of the modified severity. The system preferably further comprises a screen for displaying and presenting to a user or operator (particularly for comparison purposes) all or part of the generated and optionally received information, i.e., severity scores (assigned to the received and modified images) and the received and modified images.

[0070] The present invention will be better understood from the following detailed description taken in conjunction with the accompanying drawings. [Brief explanation of the drawings]

[0071] [Figure 1] 1 is a schematic diagram of the main steps of the method that is the subject of the present application; [Figure 2]FIG. 1 is a schematic diagram of a Bayesian network applied to determine a modified severity score of visible signs of aging. DETAILED DESCRIPTION OF THE INVENTION

[0072] As shown in FIG. 1, the purpose of the method according to the present application is to model, predict, and evaluate the progression of signs of aging in a user and, if applicable, generate a virtual image of their estimated appearance in several years.

[0073] For this purpose, the method uses an image or photograph Pi of at least one body region of the user that typically exhibits visible signs of aging, in this case, the image Pi being a photographic portrait of the user's face, including in particular the region S1 of nasolabial folds, the region S2 of crow's feet, the region S3 of wrinkles and dark circles around the eyes, and the region S4 of forehead wrinkles.

[0074] The method generates a modified image Pm of the user as a function of various aging factors fi that are likely to affect the progression of signs of aging present in the initial image Pi, including, inter alia, amount of sleep, use of cosmetics, sports activity, sun exposure, alcohol consumption, facial expression, and body mass index (all or some of these).

[0075] Thus, in a first step, a portrait photographic image of the user is taken. Advantageously, the photograph is taken using a high-definition digital camera, preferably in standard lighting conditions. Alternatively, the image can be acquired using a mobile device such as a smartphone or tablet computer with an integrated camera. The digital photograph can then be processed at a later date, particularly to correct exposure, color, etc.

[0076] The image data is then processed to identify and analyze areas of the body that typically exhibit signs of aging, with the goal of assigning an initial severity score to said signs of aging.

[0077] As mentioned above, each body region of interest can be identified using one or more computer vision algorithms, in particular artificial intelligence algorithms trained for this purpose on the basis of portrait photographs. The detection and demarcation of each region of interest can advantageously be based on the detection of body markers, in particular facial markers.

[0078] Then, a severity score N for the signs of aging considered for each body region of interest is calculated. Si The severity score is advantageously determined and assigned by an image analysis algorithm, in particular an artificial intelligence algorithm trained for this purpose.

[0079] For further details of such a method, please refer to the aforementioned application (Patent Document 1). Naturally, the method is not limited to such system implementation, and any method, particularly a computer-implemented method, for identifying and segmenting body regions may be considered. The body regions thus segmented may be scored on a severity scale by any means.

[0080] According to the method that is the subject of the present application, the severity score assigned to each sign of aging in each body area of ​​interest is modified as a function of the value given to the aging factor fi considered.

[0081] Thus, at the same time as the image data is received, values ​​are received for each of the aging factors fi to be considered, at least some of the received values ​​being adapted and personalized to the user according to the user's lifestyle habits, and these values ​​can be obtained via a questionnaire and / or can be entered manually via a computer interface.

[0082] It should be noted that not all aging factors fi are necessarily relevant to each sign of aging considered, and a progression model specific to a sign will use values ​​assigned to factors fi that are considered relevant to its progression. For example, it is known that the progression of pigmentation spots is fundamentally dependent on exposure to sunlight. Therefore, a progression model applied to assess the progression of this sign of aging will use values ​​assigned to the relevant fi factors, even if requested by the user specifically for the purpose of assessing the progression of other signs of aging, for which other factors may be relevant.

[0083] Preferably, the aging factors offer a limited number of possible values ​​(modalities). In particular, the possible values ​​represent the intensity and / or frequency of exposure to said factors. Some factors can be Boolean, i.e., have two possible values: YES / NO, 0 / 1, type, etc. Interestingly, other factors can have three or even four possible values, allowing them to take on a median or average intensity value.

[0084] According to a first embodiment, the aging factor accepts only one input selected from among possible values. According to a second embodiment, one or more aging factors accept several inputs from among proposed possible values, each input being associated with an intensity parameter or weighting factor determined for the user. This type of operation allows better consideration of intermediate states whose impact may be difficult to assess.

[0085] Thus, for example, for an aging factor such as facial expressivity, an expert could probably easily assess the impact of aging on extreme situations, i.e., over- or under-expressivity. However, "average" expressivity would be nearly impossible to assess and model. According to this embodiment, the user is asked to enter personalized weighting factors for each proposed value ("high expressivity" and "low expressivity"). Thus, instead of receiving values ​​corresponding to deterministic states, it is possible to consider weighted values.

[0086] Thus, using the example of facial expressivity, a user may consider themselves to be 30% "high expressive" and 70% "low expressive." These weighting factors are reported and used to identify the influence of the corresponding aging factors based on the established influence of the proposed deterministic extremes. The human-machine interface used to collect these values ​​and weights is configured to receive and process this data accordingly.

[0087] For example, research has shown that standard / neutral expressivity has no effect (impact equal to 0) on the acceleration of age-expressive wrinkle salience. On the other hand, research has shown that overexpressivity has a significant effect on the acceleration of age-expressive wrinkle salience (e.g., 15% chance of a 3-degree worsening in severity over 15 years, 30% chance of a 2-degree worsening, 40% chance of a 1-degree worsening, 15% no effect). Weights previously received from the user for these values ​​corresponding to the determined extremes are then used to determine a user-adapted impact probability table, as described below.

[0088] Similar implementations can be implemented for age-related factors such as sleep quality, stress levels, and exposure to pollution.

[0089] According to one embodiment, the factors fi are determined by the severity score ΔN Si The input variables (modulators) of a computer model M are configured to determine the variation of at least one of the factors f i as a function of the values ​​(evidence or observation nodes) input for these factors f i , and the determined variation is used to generate a modified severity score N' Si The initial severity score N assigned to the considered signs of aging to obtain Si All evidence forms a scenario that corresponds to the user's simulated or actual situation.

[0090] In particular, the computer model M implements a causal Bayesian network, some favorable structural features of which are provided below.

[0091] Aging factors i For each input variable corresponding to , the model calculates an influence I, which represents the variation in the severity of the considered factor. i Presents the dependency or causal link to the node that defines the impact I. i takes the form of a table of conditional probabilities relating each modality of the factors fi to the variation in the degree of progression. In particular, the probability tables forming the influence Ii of each factor can be obtained by expert knowledge and / or elicitation from experimental data.

[0092] Preferably, the Bayesian network structure includes only one level of such derived influence nodes.

[0093] The influence node Ii has a total final change ΔN Si causally linked to pooled downstream nodes Aij with the aim of grouping together probabilities and variations according to the severity of several factors until it can be determined. The factors to be grouped into pooled nodes can also be determined by elicitation, in particular by type of factor (external, internal, linked to the same external, etc.).

[0094] In particular, factors with negative impacts may be grouped separately from factors with positive impacts in one or more steps to determine the nodes.

[0095] Each pooled node Aij is annotated with a table of conditional probabilities that associate the progression variation that can result from the association of these factors. Various operations can be performed to capture the possible variations, for example, it is possible to retain the minimum variation between parents, the maximum variation between parents, or the sum of the variations between parents, with it being noted that such a sum is preferably bounded from above and / or below. Thus, for example, it can be considered that the variation resulting from smoking and sun exposure cannot exceed +2. The associated probabilities are calculated accordingly, in particular by applying Bayes' theorem. The nature of the operations to be applied can be determined by an expert.

[0096] The final pooled node BT corresponds to the final total variation applied to the initially determined severity level. This node is obtained from the progressive pooling of all considered factors and non-user-specific influence nodes T and represents residual effects or effects without known and / or quantifiable causes, especially those resulting simply from the passage of time. The action applied is the sum of the variations, which is limited or constrained by the highest variation among the aggravation modalities of the considered factors.

[0097] In addition, the sum of the fluctuations for determining the final pooled node BT is also constrained to be strictly greater than zero (negative fluctuations imply rejuvenation).

[0098] Of course, other Bayesian network structures are possible. However, such a model has the advantage that it remains easy to interpret (predictions are simply the result of initial conditions to which the influence of various factors / adjustment coefficients is added). It is also a "white box" model, i.e., the calculations are known.

[0099] FIG. 2 shows an example of an application for determining the progression of a user's nasolabial folds S1.

[0100] As mentioned above, the first step is to assign an initial severity score N to the sign of aging being considered. S1 In this case, the severity score is 1 on a scale of 0 to 5.

[0101] At the same time, the user indicates personalized values ​​for various aging factors that may exacerbate the progression of nasolabial folds.

[0102] In this case, the user indicated the following personalized values: f1 - Prolonged sun exposure (light exposure): "Medium" from the following options: "Almost never or never"; "Low"; "Medium"; "High"; f2 - Application of photoprotective products: "Regularly" from the following options: "Never / Rarely"; "Recreationally"; "Regularly"; f3 - Smoking: 10-20 packs per year, from the following options: "less than 10 packs per year"; "10-20 packs per year"; "more than 20 packs per year"; f4 - Face Shape: The following options are available: "Short and wide"; "Neutral"; "Oblong"; "Neutral"; f5 - Sleeping on your back: Choices: "Yes" or "No"; then "No"; f6 - Skin care product application: from the following options: "none"; "daily moisturizer"; "moisturizer and anti-aging"; "moisturizer and anti-aging".

[0103] A corresponding impact I1...I6 is determined for each factor. As mentioned above, this impact is determined based on expert knowledge, particularly through elicitation. Each impact indicates the probability of a given variation of a certain degree of severity, depending on the value of the factor considered.

[0104] Thus, for factor f1, which represents the user's light exposure, the highest probability is that this "average" exposure will have no significant long-term effect (zero variation). Of course, the probability of a one-degree worsening is lower, and the probability of a two-degree worsening is even lower.

[0105] A probability table is also determined for the f2 factor, which represents the application of a photoprotective product. For the user considered, the application of such a product is "regular" and therefore the most likely outcome is zero effect on the severity of nasolabial folds. As with the f1 factor, the probability of a 1-degree improvement is lower (variation -1) and an even lower probability of a 2-degree improvement.

[0106] Then, the influences I1, I2 of factors f1 and f2 represent the overall influence of the sun, and a pooled node A is constructed from a probability table for each possible variation in the degree of severity. 12 In this case, the fluctuations are added and are greater than or equal to 0. The probability of each fluctuation is calculated accordingly.

[0107] A similar procedure is used for the other factors f3 to f6: the effects I3, I4 of factors f3 and f4 are pooled together, while the effects I5, I6 of factors f5 and f6 are pooled together and are considered to be the improving factors.

[0108] Then, all pooled nodes A12, A34, A56 are pooled into a subordinate node A123456, accumulating all estimated variations for the considered factors f1 to f6.

[0109] The final variation BT is determined by applying a probability table for the time factor T, which represents the residual effect at 15 years. As shown, at 15 years, the likelihood of nasolabial fold progression is 2 degrees, slightly higher than a 1 degree variation. This correspondence table is determined by eliciting expert knowledge. Note that, in addition to the time factor T, the final variation BT can only take positive values, so only variations greater than or equal to zero are retained from the pooled nodes to accumulate all estimated variations.

[0110] Finally, the corrected severity score N' Si is a table of conditional probabilities of the variable BT with the initial severity score N Si The modified severity can be limited by a maximum severity score that cannot be exceeded (e.g., in this case, 5, i.e., the maximum value of the scoring scale). Thus, if the initial severity score is 1, the probability of the variation being 5 will not result in a severity score of 6, but rather the probability of reaching the maximum severity score, i.e., 5.

[0111] Thus, a probability distribution is obtained for different degrees of the scoring scale for the signs of aging considered.

[0112] A modified image Pm representing said signs of aging with modified severity can be generated for the most probable modified severity, in this case for example a severity of 3. As mentioned above, it is also possible to generate a modified image of average severity, obtained by weighting with the probability distribution obtained for each variation in particular initial severity.

[0113] The initial severity score of the signs of aging and the impact and pooled data on the nodes may depend on the skin type of the user, for this purpose the method may include the additional step of receiving this typology information in order to select an appropriate assessment and simulation model.

[0114] A similar approach can be applied to the user's age, with the difference being that only the statistical models and predicted progression data may be a function of the user's age range, and the assessment of the initial severity score is preferably not based on a model that is a function of the user's age. However, the user's actual age can be used to improve, enhance, or confirm the scoring model used.

[0115] Therefore, dedicated progression models can be provided for several age ranges, in particular a model for the 30-40 age range, a model for the 40-50 age range, etc. It would then be advantageous to implement an "age continuous" model via an interpolation function that allows accepting any age value, for example, from 18 to 65, without having to resort to an age range model.

[0116] More specifically, an interpolation function is provided for weighting the results obtained using one or more models. For example, for a 25-year-old user, it would be possible to use a model designed for the age range 18-35 years and weight the predictions by a model designed for a target age, i.e., 40 years (25 years + 15 years), i.e., for example, a model for the age range 35-50 years.

[0117] In fact, the rate of aging is slower for those aged 18–35 than for those aged 35–50. Using only the 18–35 model may result in an underestimation of aging, whereas using only the 35–50 model risks overestimating aging.

[0118] One solution is to apply both models to the user and take a weighted combination of their results, where the weights can be determined by an aging function.

Claims

1. 1. A computer-implemented method for modeling and assessing the progression of signs of aging (S1, S2, S3, S4) in a user, comprising: receiving data corresponding to a set of pixels of an image (Pi) of a body region of the user generally exhibiting at least one visible sign of aging; Severity score (Ns ｉ ) to visible signs of aging of said body region; Aging factors (f ｉ receiving data corresponding to personalized values ​​adapted to said user for all or some of the items; At least one aging factor (f ｉ , T) as a function of the value of the severity score (N's ｉ ) and Including, The method is characterized in that the severity score is modified by adding a progressive variation (ΔNsi) to an initial severity score (Nsi), the progressive variation being determined from the value of the aging factor (fi).

2. 2. The method of claim 1, further comprising the additional step of generating a modified image (Pm) of the body area of ​​the user having visible signs of aging corresponding to the modified severity score.

3. 3. The method according to claim 1 or 2, characterized in that at least one ageing factor (T) is independent of the user.

4. The progress variation (ΔNs ｉ ) is a method for calculating at least one probability model (M) based on the aging factor (f ｉ 4. The method according to claim 1, wherein the value of the parameter is determined by applying the parameter to the value of the parameter.

5. The method according to claim 4, characterized in that the probabilistic model (M) is a causal Bayesian network.

6. at least one first data input configured to receive data corresponding to a set of pixels of a digital image (Pi); ｉ a second data input configured to receive at least one value of Based on the received pixels, a severity score (NS) of at least one visible sign of aging (S1, S2, S3, S4) is determined. ｉ ) and modifying the severity score as a function of the received value of the aging factor; The corrected severity score (N's ｉ ) and A system for implementing the method according to any one of claims 1 to 5, characterized in that it comprises at least one processor configured to:

7. 7. The system of claim 6, further comprising a digital photo capture device in communication with the first data input.

8. From the received image data (Pi), the corrected severity score (N's ｉ 8. A system according to claim 6 or 7, characterized in that it comprises an image generator (GAN) adapted to generate a modified image (Pm) representative of the image.

Images