JP7407115B2 - Machine performing facial health and beauty assistant - Google Patents

Description

CROSS-REFERENCE TO RELATED APPLICATIONS This application refers to U.S. Provisional Patent Application No. 62/614,001, both filed January 5, 2018, and U.S. Relating to and claiming the benefit of Provisional Patent Application No. 62/614,080.

The health and beauty industry takes advantage of advances in technology that improve the consumer experience with their products and services. Certain websites now make use of facial recognition technology that identifies human facial features (eyes, cheeks, nose, lips, chin, etc.) provided through mobile devices, for example. Such computer vision technologies have not been able to embrace the full power of machine learning, especially where customization to specific consumers is involved. That is, for one, traditional health and beauty portals lack a mechanism whereby regimens are recommended by a machine, as opposed to being recommended by a human clinician. In embodiments, a regimen is a systematic plan or course of action intended to improve the health and/or beauty of a human user. In the area of facial health and beauty, regimens include cleaning the skin with certain cleansers and applying certain creams, observing certain dietary restrictions, changing sleeping habits, etc. Sometimes.

Traditional health and beauty portals also lack a mechanism by which a user's skin can be tracked over time, such as to monitor the effectiveness of recommended regimens. They lack well-informed explanations and advice, are not targeted separately for men and women, and suffer from false positives when detecting certain conditions (e.g. hair misidentified as wrinkles). . The ability to recommend a health and/or beauty regimen through data analysis and to track an individual user's progress through that regimen through data analysis has not yet been realized on a machine.

One or more images representing the user's facial skin are received from the user by one or more processing circuits. In embodiments, machine learning models stored in one or more memory circuits are applied to images to classify facial skin characteristics, identify objects of interest, determine cosmetic trends, etc. In embodiments, regimen recommendations are provided to the user based on the classified facial skin characteristics.

In embodiments, one or more memory circuits configured to store a machine learning model and at least one image representative of facial skin of a user are received from a user, and the machine learning model is configured to store the machine learning model in the image to classify facial skin characteristics. and one or more processing circuits configured to generate regimen recommendations to a user based on the classified facial skin characteristics.

In embodiments, the one or more processing circuits receive from a user another image representing facial skin of the user, apply a machine learning model to the other image to classify facial skin characteristics, and apply a machine learning model to the other image to classify facial skin characteristics. The method is further configured to update regimen recommendations to the user based on the determined facial skin characteristics.

In embodiments, the one or more processing circuits are further configured to process the image such that the user's facial skin progresses to a simulated future state.

In embodiments, the simulated future state is a simulated future state of the user's facial skin when the regimen is followed by the user.

In embodiments, one or more processing circuits are physically separated into a client platform and a service platform that are communicatively coupled through a communications network.

In embodiments, the one or more processing circuits are further configured to accept input from a plurality of users to classify facial skin characteristics from training images provided to it and to train a model using the received inputs. ing.

Embodiments include receiving from a user at least one image representative of the user's facial skin, applying a machine learning model to the image to classify facial skin characteristics, and generating a user image based on the classified facial skin characteristics. A method is provided that includes: generating a regimen recommendation for a patient.

In embodiments, a processing circuit receives at least one image representing facial skin of a user, a communication circuit communicates the received image to a machine learning model and receives a regimen recommendation from the machine learning model, and a communication circuit receives the regimen recommendation from the user. An apparatus is provided that includes a user interface circuit for presenting a user interface circuit.

In embodiments, the processing circuitry is configured to alert the user that another image representing the user's facial skin is required according to a predefined schedule, and to accept from the user another image representing the user's facial skin. further configured to communicate the another image to the machine learning model and receive an updated regimen recommendation from the machine learning model, the user interface circuit further configured to communicate the updated regimen recommendation to the machine learning model; is further configured to present to the user.

In embodiments, the user interface circuit is further configured to present an image of a human face to the user, and the processing circuit receives input from the user, through the user interface circuit, to classify facial skin characteristics from the image provided to it. The communication interface circuit is further configured to receive, and the communication interface circuit is further configured to communicate user input as training data to the machine learning model.

In embodiments, the user interface circuit is further configured to present user controls by which facial skin characteristics are evaluated on a predetermined scale.

Embodiments further include a camera communicatively coupled to the processing circuit to provide images thereto from a user.

In embodiments, the camera, processing circuitry, user interface circuitry, and communication circuitry are components of a smartphone.

Embodiments include receiving at least one image representative of facial skin of a user, communicating the received image to a machine learning model, receiving regimen recommendations from the machine learning model, and transmitting regimen recommendations to the user. A method is provided that includes:

1 is a schematic block diagram of an exemplary system configuration by which the present general inventive concept may be embodied; FIG. 3 is a flowchart of a simple user interaction with an embodiment of the present general inventive concept; 1 is a schematic block diagram of data flow for an example embodiment of the present general inventive concept; FIG. 1 is a block diagram of crowdsourced training of machine learning models in accordance with an embodiment of the present general inventive concept; FIG. 1 is a diagram of an example client platform device upon which the present general inventive concepts may be implemented; FIG. 2 is a flowchart of crowdsourced training of an example machine learning model in accordance with an embodiment of the present general inventive concept; FIG. 3 is a diagram illustrating a test operation according to crowdsourced machine learning model training.

The inventive concepts are best explained through the specific embodiments thereof, which are herein described with reference to the accompanying drawings, in which like reference numerals refer to like features throughout. The term invention, as used herein, is intended to allude to the inventive concept underlying the embodiments described below, and not merely to the embodiments themselves. It will be understood that this is not the case. Furthermore, it will be understood that the general inventive concept is not limited to the exemplary embodiments described below, and the following description should be read with such an understanding.

Additionally, the word exemplary is used herein to mean "serving as an example, instance or illustration." Any embodiment of a structure, process, design, technique, etc. specified herein as exemplary is not necessarily to be construed as preferred or advantageous over other such embodiments. No particular quality or suitability of the embodiments shown herein as illustrative is intended or assumed.

DESCRIPTION FIG. 1 is a schematic block diagram of an exemplary facial health and beauty assistant (FHBA) system 100 that includes an FHBA client platform 110 and a FHBA service platform 120 communicatively coupled through a network 130. . In one embodiment, FHBA client platform 110 is a smartphone, tablet computer, or other mobile computing device, although the invention is not limited thereto. As illustrated in FIG. 1, an example FHBA client platform 110 includes a processor 112, a memory 114, a camera 115, a user interface 116, and a communication interface 118 on which an FHBA client interface 150 may be implemented. including. FHBA client interface 150 provides a primary portal through which users access FHBA system 100.

In one embodiment of the invention, FHBA service platform 120 includes one or more server computers, each including a processor 122, memory 124, user interface 126, and communication interface. Those resources of FHBA service platform 120 may be utilized to implement FHBA service interface 152, machine learning logic 154, and storage memory 156. Storage memory 156 represents a sufficient amount of volatile and persistent memory to embody the invention. Storage memory 156 may further contain space for a single user's private profile, as well as a vast amount of encoded human knowledge. Storage memory 156 may further store processor instructions that, when executed by one or more processors 122, perform a number of tasks or procedures for embodiments of the invention. Storage memory 156 further includes a user model (coefficients, weights, processor instructions, etc.) operable by machine learning logic 154 to define a particular regimen for the user and track the user's progress under the regimen. may be memorized.

The example FHBA service interface 152 provides an infrastructure by which network access to FHBA services is both facilitated and controlled. FHBA client interface 150 and FHBA service interface 152 communicate over appropriate communications links 145 using signaling and data transmission protocols to which communications interfaces 118 and 128 are constructed or otherwise configured. to communicate. FHBA service interface 156, along with appropriate Internet hosting services, may implement authentication and other security mechanisms to allow access only to authorized users and protect users' private data. Additionally, FHBA service interface 152 may implement an application programming interface (API) that enables FHBA client interface 150 to communicate with machine learning logic 154, for example. Those skilled in the art will recognize other front end services that can be used with the present invention.

Machine learning logic 154 provides an infrastructure for embodiments of the invention to learn from and make predictions about data without being explicitly programmed to do so. In certain embodiments, machine learning logic 154 includes one or more convolutional neural networks (CNNs), models that can be trained using open source datasets, or crowdsourced models, as described below. Implement the dataset. including but not limited to decision tree learning, association rule learning, artificial neural networks, deep learning, inductive logic programming, support vector machines, clustering, Bayesian networks, reinforcement learning, representation learning, similarity and distance learning, sparse dictionary learning, Other machine learning techniques may be used with the present invention, including genetic algorithms, rule-based machine learning, and learning classifiers. U.S. Patent No. 8,442,321, U.S. Patent No. 9,015,083, U.S. Patent No. 9,536,293, U.S. Patent No. 9,324, all of which are incorporated herein by reference. Additional techniques described in No. 022 and US Patent Application Publication No. 2014/0376819 may be used with the present invention. In the description that follows, it is assumed that the machine learning logic implements a convolutional neural network, but the invention is not so limited. Those skilled in the art of artificial intelligence will recognize numerous techniques that can be used with the present invention without departing from its spirit and intended scope.

Embodiments of the invention determine various regimens for the user based on images of the user taken by camera 116 on FHBA client platform 110. In certain embodiments, images of the user's face are preferably obtained under conditions of uniform illumination that is consistent over time. To this end, and with reference to FIG. 1, embodiments of the invention provide a mirror device 140 that includes a mirror surface 144 surrounded by an annular illuminator 142. This configuration is intended to define a temporally constant reference of illumination. When the invention is embodied in this manner, time-varying characteristics in images of a user's face are more easily recognized and labeled.

FIG. 2 is a flowchart by which example interactions may be described using embodiments of the invention. The interaction of FIG. 2 is simple in design and is not intended to be limiting in any way. The description of FIG. 2 is intended to illustrate the functionality of the configuration illustrated in FIG. Beyond the features described with reference to FIG. 2, further features of the invention are discussed below.

In act 210, the user may generate an image of the user's face, such as by camera 116 of FHBA client platform 110. This may be accomplished with or without the lighting criteria discussed above. In act 215, the user's image is sent to the FHBA service platform 120. This may be accomplished by a suitable communication protocol shared between FHBA client platform 110 and FHBA service platform 120 to implement communication link 145.

In operation 220, image analysis and machine learning is performed to analyze the user's skin from the image. The machine learning logic 154 includes, among other things, visual age, i.e., the user's subjective age as estimated from the visual appearance of the user's face, the uniformity of facial skin tone (bumps, age spots/sun spots), , acne scars, and other imperfections), puffiness under the eyes, dark circles around the eyes, overall drooping tone at the eyelids/corners of the mouth, fine lines, and the presence of stress as seen in the redness of the eyes, flakiness , hydration level, often referred to as plump or smooth, indicating the lack of peeling, dullness, and fine lines in the skin, radiance – a non-linear parameter where the ideal is a moderate amount of radiance, pore condition – fewer pores Presence of acne, wrinkles, folds in the skin, which can be discovered through analysis, as the appearance is characterized by red/red swollen pimples and blemishes, which is desirable as it gives a healthy, youthful and flat skin feel. Analysis to determine the presence of ridges or folds, the presence of sagging, i.e., the drooping appearance of soft tissues caused by decreased elasticity, as well as the presence of crow's feet, branching wrinkles, particularly located around the periphery of the human eye. may be executed. Other conditions of the skin may be determined by machine learning logic 154. Further details of the analysis are provided below. Once the analysis is complete, process 200 provides the analysis results and the prescribed regimen (products and routines) and/or updates to the regimen to the user via FHBA client interface 150, as determined in operation 225. The information may be transmitted to act 220.

In act 225, it is determined whether the analysis is complete, and in response to the affirmative determination, process 200 causes FHBA service interface 152 to send the recommended regimen or updates to the regimen to FHBA client interface in act 230. 150, and may transition to act 230. The user may follow a regimen as shown in act 235, and in act 240 it is determined whether a new interval has begun. If so, process 200 repeats from operation 210. FHBA client interface 150 has access to a calendar and timer (also GPS) onboard FHBA client platform 110 and may access a network-accessible calendar over network 130. Thus, for example, once a week, the FHBA client interface 150 may remind the user to take a picture of the user's face, ie, remind the user of a new interval. Over time, the FHBA system 100 can determine from images taken at each interval whether the recommended regimen is working, and if not, the FHBA system 100 can revise the regimen. For example, it may change products, recommend further lifestyle changes, make doctor's appointments, etc.

FIG. 3 is a diagram of data flow between an example FHBA client interface 150 and services of FHBA service platform 120. It should be noted that in FIG. 3, the FHBA service interface 152 has been omitted to avoid unnecessary clutter in the diagram. However, those skilled in the relevant art will recognize the operation of the FHBA service interface 152 that controls and facilitates the data flow depicted in FIG.

As illustrated in FIG. 3, machine learning logic 154 may include a skin analyzer 330, a facial appearance progression generator 335, and a regimen recommendation generator 340, and is communicatively coupled to a user account database 310 and a product database 320. You can. Machine learning logic 154 may train and utilize machine learning model 370 to recommend regimens and track the user's progress through the regimen. As those skilled in machine learning will attest, training involves selecting a set of features, e.g. appearance age, uniformity, stress, hydration, shine, pores, acne, wrinkles, sagging, crow's feet, etc. , and assigning labels to the image data that reflect the presence or prominence of those features. Assigning labels may be performed by subject matter experts or through crowd-sourced data as described below. Considering the assigned label as the correct answer, machine learning logic 154 may configure model 370 to predict the degree to which the feature is present in the test image, which may change over time. The invention is not limited to particular model representations, which may include binary models, multi-class classification models, regression models, and the like.

The exemplary user account database 310 contains data for all users of the FHBA system 100 in a secure manner. This includes user profile data, current and past user photos 357 for each user, current and past skin analysis 358 for each user, and current and past product recommendations 362 for each user. routine recommendations 364.

Exemplary product database 320 includes data for different products that can be used in a regimen. Product database 320 may include records reflecting product name, active and inactive ingredients, label information, recommended uses, and the like. In certain embodiments, users (and other users of FHBA system 100) may provide feedback regarding different products, as illustrated by product input 354, and may enter products that are not present in product database 320. You may. The present invention is not limited to specific products that can be entered into product database 320.

Skin analyzer 330 is constructed or otherwise configured to classify various skin conditions or artifacts from the image of the user's face using machine learning techniques through model 370. In certain embodiments, a photographic image 352 of the user's face is provided to the skin analyzer 330 for analysis. Skin analyzer 330 may implement image pre-processing mechanisms including cropping, rotating, registering, and filtering the input image prior to analysis. After any such pre-processing, skin analyzer 330 may apply model 370 to input image 357 to identify, identify, and classify characteristics of the user's facial skin.

Facial appearance progression generator 335 may operate on the user's facial image to depict what the user's face will look like at some time in the future. Such progression may be due to age, for which age progression techniques can be deployed, or to appearance resulting from compliance with a regimen. The advanced image 356 may be provided to the user through the FHBA client interface 150.

The regimen recommendation generator 340 may operate on the analysis results 358 obtained from the skin analyzer 430 to the user toward prescribing a regimen. The model 370 may be trained to predict which products and routines (such as treatments, cosmetics and lifestyle recommendations) will be effective in meeting the user's goals with respect to facial skin characteristics identified in the skin analysis. good. The regimen recommendation generator 340 may format the analysis results 358 of the skin analyzer 330 as a query to the product database 320, for example, based on the knowledge encoded for the model 370. In response, product database 320 may return product data and metadata 366, and product recommendations 362 and routine recommendations 364 may be provided to FHBA client interface 150.

As indicated above, training of model 370 may be accomplished by labeling image data by an expert. However, instead of experts, certain embodiments of the invention utilize crowdsourced data as training data. FIG. 4 is a diagram of such an embodiment of the invention. During training, the user 410 is presented with a set of training images 420 and asked to characterize facial skin properties and/or facial features for the set of training images 420. In one embodiment, a suitable scale (eg, an integer from 1 to 10) is constructed by which a user can rate the severity or salience of a feature. For example, each of the users 410 is presented (over time) with a number of facial images and walked through a set of questions regarding the features and/or skin characteristics of the person in the images. Using a scale (1-10), each user 410 can determine each of the characteristics (e.g., appearance age, evenness, stress, hydration, shine, pores, acne, wrinkles, sagging, crow's feet, etc.) asked to rate the prominence of Answers to questions may serve as labels used to train machine language logic 154.

Referring to FIG. 5, an exemplary FHBA client platform 110 in the form of a smartphone having a touch screen 510 as the user interface 118 is illustrated. The example FHBA client interface 150 is implemented on the computational resources of the FHBA client platform 110 as discussed with reference to FIG. The FHBA client interface 150 may present a photograph of a human face in the image area 520 and, via text 142, ask, ``On a scale of 1 to 10, how would you rate the presence of crow's feet on this person?'' "1" means "not visible," and "10" means "prominently present." A suitable user interface control 144 (a slider control illustrated in FIG. 5) may be implemented on the FHBA client interface 150 to allow the user to enter his or her rating.

FIG. 6 is a flowchart of a crowdsourced training process 600 by which the present invention may be implemented. At operation 610, training images may be provided to the FHBA client interface 150. The set of training images 420 may be preselected as including exemplary examples of skin characteristics of interest. In act 620, the user is provided with a first question and waits for an answer (rating) in act 630. Such an answer could be, for example, ``On a scale of 1 to 10, how would you rate this model's acne?'' ``1'' is ``not visible,'' and ``10'' is ``very severe.'' It may also be 'prominent'. When the user answers the question, as determined in operation 630, the user's answer may be formatted into a label appropriate for machine training of machine learning logic 154 in operation 640. In operation 650, it is determined whether all questions associated with the currently displayed image have been answered. If not, the process 600 transitions again to operation 620 where the next question is posed. If all questions have been answered, as determined in act 650, it is determined in act 560 whether all training images have been presented. If not, the process 600 transitions again to operation 610 where the next training image is presented. If all training images have been presented, as determined in operation 660, the labeled images may be used to train model 370 in operation 670.

All iterations in process 600, such as presenting the next question in acts 620 and 650 and/or presenting the next image in acts 610 and 660, need to be performed in any one sitting. There isn't. For example, a user may be prompted to answer a single question at once (e.g., each time the user logs on), and only over time will all questions and images be answered by any one user. Presented. Alternatively, the user may choose to answer all questions for all images in a single sitting. Many labels may be generated for the training model 370 over a large number of user and/or facial images, and the statistical trends underlying such training are different from those of human experts. On the contrary, it reflects the public view.

FIG. 7 illustrates the testing operation of an embodiment according to the crowdsourced training discussed above. Test images 710, ie, the user's own images, may be presented to machine learning logic 154, which analyzes the images for each model trained against crowdsourced data 720. As illustrated in the figure, the machine learning logic 154 determines that 80% of those surveyed rate the user's crow's feet 7 out of 10 in terms of prominence, as shown at 722. It is estimated that the evaluation is as follows. Accordingly, machine learning logic 154 may recommend a regimen (eg, a cream specifically formulated for crow's feet and recommend application instructions) based on a severity score of 7.

As will be appreciated by those skilled in the art, aspects of the invention may be embodied as a system, method, or computer program product. Accordingly, aspects of the invention may be implemented in an entirely hardware embodiment, an entirely software embodiment, all of which may be referred to herein as a "circuit," "module," or "system." It may take the form of an embodiment (including firmware, resident software, microcode, etc.) or an embodiment combining software and hardware aspects. Furthermore, aspects of the invention may take the form of a computer program product embodied in one or more computer readable medium(s) having computer readable program code embodied thereon.

Any combination of one or more computer readable media(s) may be utilized. The computer readable medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include an electrical connection having one or more wires, a portable computer diskette, a hard disk, a solid-state disk, a random access memory (RAM), a read-only memory ( ROM), erasable programmable read-only memory (EPROM or flash memory), fiber optics, portable compact disk read-only memory (CD-ROM), optical storage, magnetic storage, phase change memory storage, or any of the above. including any appropriate combination of the following: In the context of this specification, a computer-readable storage medium is any tangible medium capable of containing or storing a program for use by or in connection with an instruction execution system, apparatus, or device. good.

Computer program code implementing operations for aspects of the invention may be implemented in one or more programming languages, such as an object-oriented programming language such as Java, Smalltalk, or C++, or the "C" programming language or similar programming language. may be written in any combination of traditional procedural programming languages. The program code may be executed entirely on a user's computer, in part on a user's computer, as a stand-alone software package, partially on a user's computer and partially on a remote computer, or on a remote computer or server. It may be performed in its entirety. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or wide area network (WAN), or a connection to an external computer may be made. (e.g., over the Internet using an Internet service provider). It will be appreciated that software for the computer system of embodiments of the invention may be developed by one skilled in the computer arts based on the functional descriptions included in the specification and the flowcharts illustrated in the figures. Furthermore, any references herein to software performing various functions refer generally to a computer system or processor performing those functions under software control.

Computers of embodiments of the invention may alternatively be implemented with any type of hardware and/or other processing circuitry. The various functions of a computer system may be distributed in any manner between any number of software modules or units, processing or computer systems and/or circuits, and the computers or processing systems may be local or It may be located remotely and may communicate via any suitable communication medium (eg, LAN, WAN, intranet, Internet, hardwired, modem connection, wireless, etc.).

The above examples are illustrative of particular functionality of embodiments of the invention and are not intended to be limiting. Indeed, other functionality and other possible use cases will be apparent to those skilled in the art upon reviewing this disclosure.

Claims (13)

one or more memory circuits configured to store the machine learning model;
receiving from the user at least one image representative of the user's facial skin;
applying the machine learning model to the image to classify facial skin characteristics;
generating regimen recommendations including product recommendations to the user based on the classified facial skin characteristics;
one or more processing circuits configured to;
a product database containing data of different products included in the regimen recommendation;
Equipped with
the one or more processing circuits are further configured to process the image using an age progression technique such that the user's facial skin progresses to a simulated future state;
The one or more processing circuits include:
alerting the user that another image representing the user's facial skin is required according to a predefined schedule;
receiving from the user another image representing the user's facial skin;
applying the machine learning model to the other image to classify facial skin characteristics;
further configured to update the regimen recommendation to the user based on the classified facial skin characteristics of the another image;
The product database is configured to receive data on products that do not exist in the product database from a user,
The system wherein the simulated future state is a simulated future state of the user's facial skin when the regimen is adhered to by the user. 2. The system of claim 1, wherein the one or more processing circuits are physically separated into a client platform and a service platform that are communicatively coupled through a communications network. The one or more processing circuits include:
receiving input from a plurality of users for classifying the facial skin characteristics from provided training images;
training the machine learning model using the received input;
The system of claim 1, further configured to. receiving from the user at least one image representative of the user's facial skin;
accepting data from a user for products that are not present in a product database that includes data for different products included in the regimen recommendation;
applying a machine learning model to the image to classify facial skin characteristics;
generating the regimen recommendations including product recommendations to the user based on the classified facial skin characteristics;
processing the image using an age progression technique such that the user's facial skin progresses to a simulated future state;
alerting the user that another image representing the user's facial skin is required according to a predefined schedule;
receiving from the user another image representing the user's facial skin;
applying the machine learning model to the other image to classify facial skin characteristics;
updating the regimen recommendation to the user based on the classified facial skin characteristics of the another image;
including;
The method, wherein the simulated future state is a simulated future state of the user's facial skin when the regimen is adhered to by the user. receiving input from a plurality of users for classifying the facial skin characteristics from provided training images;
training the machine learning model using the received input;
5. The method of claim 4, further comprising: a processing circuit that receives at least one image representing facial skin of a user;
a communication circuit that transmits the received image to a machine learning model and receives regimen recommendations including product recommendations from the machine learning model;
a user interface circuit for presenting the regimen recommendation to the user;
a product database containing data of different products included in the regimen recommendation;
Equipped with
The processing circuit includes:
alerting the user that another image representing the user's facial skin is required according to a predefined schedule;
further configured to receive from a user the another image representing facial skin of the user;
The communication circuit communicates the another image to the machine learning model and receives an updated regimen recommendation from the machine learning model applied to the another image if the regimen recommendation is not working. further configured to receive;
The user interface circuit is further configured to present the updated regimen recommendation to the user;
The product database is configured to receive data from a user on products that do not exist in the product database. the user interface circuit is further configured to present an image of a human face to the user;
The processing circuit is further configured to receive input from the user through the user interface circuit to classify facial skin characteristics from the image provided;
The communication circuit is further configured to communicate the user input as training data to the machine learning model.
7. Apparatus according to claim 6. 8. The apparatus of claim 7, wherein the user interface circuit is further configured to present a user control in which the facial skin characteristic is evaluated on a predetermined scale. 7. The apparatus of claim 6, further comprising a camera communicatively coupled to the processing circuit such that the image is provided from the user to the processing circuit. 10. The apparatus of claim 9, wherein the camera, the processing circuit, the user interface circuit, and the communication circuit are components of a smartphone. accepting at least one image representing facial skin of a user;
accepting data from a user for products that are not present in a product database that includes data for different products included in the regimen recommendation;
transmitting the received image to a machine learning model;
receiving the regimen recommendations including product recommendations from the machine learning model;
presenting the regimen recommendation to the user;
alerting the user that another image representing the user's facial skin is required according to a predefined schedule;
receiving from a user the another image representing facial skin of the user;
communicating the another image to the machine learning model;
updating the regimen recommendation if the regimen recommendation is not working;
receiving an updated regimen recommendation from the machine learning model applied to the another image;
presenting the updated regimen recommendation to the user;
including methods. presenting an image of a human face to the user;
accepting input from the user to classify facial skin characteristics from the provided image;
communicating the user input as training data to the machine learning model;
12. The method of claim 11, further comprising: 13. The method of claim 12, further comprising providing user control over which the facial skin characteristics are evaluated on a predetermined scale.

Images