KR20210143579A - Curation based cosmetics recommendation system and method for the system - Google Patents

Abstract

The present invention relates to a curation based customized cosmetics recommendation system. A cosmetics recommendation system according to the present invention includes: a response information collecting unit collecting user response information for each step of a question in a formatted form from an online platform; a request information receiving unit receiving user request information in a personalized form including image data from a user; and a cosmetics recommendation unit generating recommended cosmetics information with reference to a table, defining recommended cosmetics for each user response, according to the user response while verifying the recommended cosmetics information in consideration of output information of a cosmetics recommendation neural network for the user request information. According to the present invention, personalized customized cosmetics can be recommended. In addition, more preferred customized cosmetics can be found and provided by referring to user's skin condition information and actual purchase history information from various cosmetics online platforms.

Description

Curation based cosmetics recommendation system and method for the system}

The present invention relates to a cosmetic recommendation method and to a curation-based customized cosmetic recommendation system.

Humans have various skin types according to genetic factors, and undergo various changes in skin types according to environmental factors. In addition, according to the trend that emphasizes taste, the skin condition required for each individual is also diversifying.

Furthermore, with the development of sensing technology, technologies are being developed as auxiliary elements to diagnose individual skin conditions through various devices and to assist in skin diseases or other professional judgments.

Therefore, in the recent beauty market, rather than recommending standardized cosmetics, it is possible to recommend personalized and customized cosmetics that can improve the skin condition while respecting individuality by considering personal information along with the diagnosis of the individual skin condition. There is a need.

However, a method for recognizing an individual's changing skin condition through self-diagnosis and curating and providing cosmetic information suitable for the image pursued by each user has not been embodied at present. The prior art (Publication No. 10-2013-0057148, method and apparatus for providing cosmetic information according to a user's skin color) only identifies skin color through a kiosk or terminal and provides cosmetic information suitable for this, and specific user requirements are reflected can't do it

An object of the present invention is to propose a system for recommending personalized customized cosmetics.

It is a further object of the present invention to propose a system for identifying both direct response information to a user's query and inherent pursuit image information, and recommending cosmetics accordingly.

An object of the present invention is to propose a system for recommending more suitable cosmetics according to the user's history information.

A curation-based cosmetic recommendation system according to an embodiment of the present invention for solving the above technical problem includes a response information collecting unit for collecting user response information for each step of a standardized form from an online platform; a request information receiving unit for receiving user request information in a personalized form including image data from the user; and a cosmetic recommendation unit that generates recommended cosmetic information with reference to a table defining recommended cosmetics for each user response according to the user response, and verifies the recommended cosmetic information in consideration of output information of a cosmetic recommendation neural network for the user request information. include

The neural network determines a difference in an external element between the user's first image data and the second image data of the request model, and the cosmetic recommendation unit verifies the recommended cosmetic information according to the cosmetic ingredient information associated with the difference in the external element. do.

The external element includes at least one element of skin tone, oiliness or dryness, pores, elasticity, and wrinkle-related information.

The response information collecting unit extracts response change information from the user's response information collected for a plurality of inquiries, and the cosmetic recommendation unit generates the recommended cosmetic information by reflecting the response change information.

The cosmetic recommendation system further includes a purchase history collecting unit that receives the user's actual purchase information from the online platform, and the cosmetic recommendation unit generates the recommended cosmetic information by reflecting the actual purchase information.

The purchase history collecting unit determines the difference in actual purchase information for each user according to the same recommended cosmetic information, and provides it to the online platform.

According to the present invention, personalized customized cosmetics can be recommended. In addition, the user's skin condition information and actual purchase history information from various cosmetic online platforms can be used to find and provide more preferred customized cosmetics.

In addition, an image that the user wants to pursue is additionally input, and the image sought by the user is compared with the state of the wise user to recommend and provide cosmetics that are helpful for improvement.

This will ultimately make it easier for users to select recommended cosmetics and to receive direct help in skin care through the selected cosmetics.

1 is a diagram showing the configuration of a cosmetic recommendation system according to an embodiment of the present invention.
2 is a diagram illustrating a detailed configuration of a cosmetic recommendation system according to an embodiment of the present invention.
3 and 4 are diagrams illustrating examples of use of the cosmetic recommendation system according to an embodiment of the present invention.
5 and 6 are diagrams illustrating a neural network of a cosmetic recommendation system according to an embodiment of the present invention.
7 is a diagram illustrating a recommendation method of a cosmetic recommendation system according to an embodiment of the present invention.

The following is merely illustrative of the principles of the invention. Therefore, those skilled in the art can devise various devices that, although not explicitly described or shown herein, embody the principles of the invention and are included in the spirit and scope of the invention. In addition, it should be understood that all conditional terms and examples listed herein are, in principle, expressly intended only for the purpose of understanding the inventive concept, and are not limited to the specifically enumerated embodiments and states. .

The above-described objects, features and advantages will become more apparent through the following detailed description in relation to the accompanying drawings, and accordingly, those of ordinary skill in the art to which the invention pertains will be able to easily practice the technical idea of the invention. .

In addition, in the description of the invention, if it is determined that a detailed description of a known technology related to the invention may unnecessarily obscure the gist of the invention, the detailed description thereof will be omitted. Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 shows the configuration of a curation-based cosmetic recommendation system 100 according to an embodiment of the present invention.

The cosmetic recommendation system 100 according to the present embodiment may receive request information including skin condition information of the user 10 and a requirement to be improved from the online platform 200 through which cosmetics can be purchased.

In order to obtain self-diagnosis information according to the current skin condition, an inquiry according to a standardized sequence may be performed on the online platform 200 , and the cosmetic recommendation system 100 may collect response information to the inquiry.

In addition, in the present embodiment, external image information sought by the user 10 may be additionally input.

For example, the user 10 may input image data of a third party as a model to resemble his/her face image data.

Specifically, one's own face image data is taken through the camera of the user terminal 300, such as a smart phone or tablet pc of the user 10, and a third-party face image such as a celebrity whose image he or she wants to resemble is stored on a website or You can download it through social networks or capture it and upload it.

Accordingly, the response information of the user 10 input through the query has a standardized data form, and the requested information input through upload is personalized data in various forms according to the user 10's shooting state and the image state of the model. can have a form.

The input information may be transmitted to the cosmetic recommendation system 100 according to the present embodiment through the online platform 200 .

The cosmetic recommendation system 100 may specifically provide cosmetic recommendation information according to the response information of the user 10 .

In addition, it is also possible to verify and select a specific cosmetic from among the recommended information by using a cosmetic ingredient having a function necessary for improvement according to the requested information of the user 10 .

Finally, the verified cosmetic recommendation information is transmitted back to the online platform 200 , and the user 10 can purchase the recommended cosmetics or refer to the selection of other cosmetics.

The cosmetic recommendation system 100 according to the present embodiment recognizes the response information and the implied request information of the user 10 based on curation, and through this, the consumer browses, selects, and purchases numerous products more simply.

At this time, the data used for purchase and the cosmetic recommendation functions are linked through a deep neural network (DNN) and a customized recommendation service is provided.

Hereinafter, the cosmetic recommendation system 100 according to the present embodiment will be described in more detail.

The cosmetic recommendation system 100 includes a response information collecting unit 110 , a request information receiving unit 120 , and a cosmetic recommendation unit 130 .

First, the response information collecting unit 110 collects step-by-step user 10 response information received from a standardized query from the online platform 200 .

Specifically, referring to FIG. 3 , the response information collecting unit 110 may provide a pop-up questionnaire 302 through the terminal 300 of the user connected to the various cosmetic online platforms 200 , and the user 10 ) can guide you to enter information about your skin type, the formulation of your preferred cosmetics, etc.

In this case, the collected response information is an answer to a question according to a predetermined order, and the answer is collected in a multiple-choice format, so it can be standardized.

Therefore, the cosmetic recommendation system 100 according to the present embodiment generates the most suitable cosmetic recommendation information by referring to the user 10's response information as standardized answer data and a recommendation table in the form of a look-up table (LUT). can do.

However, the disadvantage of such structured data is that since the data itself is standardized, personalized requirements cannot be freely input, so the cosmetic recommendation system 100 collects additional information.

Accordingly, the request information receiving unit 120 receives the user 10 request information in a personalized form including image data from the user 10 .

Referring to FIG. 4 , the request information receiving unit !20 may receive image data uploaded by the user 10 using a user terminal 300 such as a smart phone.

In this case, the image data may include image data 304 obtained by directly photographing the face of the user 10 and image data 306 of a model having an image desired by the user 10 .

That is, the user 10 can select and upload images of various models existing on the web according to the image he or she desires, and allows the input of information freely in a format.

The cosmetic recommendation system 100 may use a difference in input image data to determine a part to be improved so as to have an image that is actually requested, and a neural network may be used for this determination.

The cosmetic recommendation unit 130 generates recommended cosmetic information by referring to the recommendation table 132 that defines the recommended cosmetics for each response of the user 10 according to the response of the user 10 , but the user's 10 personalized request. The recommended cosmetic information may be additionally verified by considering output information of the cosmetic recommendation neural network 134 for the information.

The neural network 134 at this time will be described in more detail with reference to FIG. 5 . In this embodiment, the neural network may be composed of an input layer, an output layer, and a hidden layer for performing calculations like a general deep learning model.

The input layer extracts feature values for input information, and the collected feature values may be input for calculation. In this embodiment, the input layer may receive the image data 304 uploaded according to FIG. 4 .

Values input to the input layer pass through each layer inside the hidden layer and are used for matrix operation.

For example, in the present embodiment, the neural network 134 is configured in the form of a Convolution Neural Network (CNN) to perform iterative convolution operation in an internal hidden layer.

Through a filter of a size determined for each layer, feature values are convolutioned to form a new matrix, and the final output value can be output through an activation function and a fully connected layer.

In this embodiment, the neural network 134 may output a skin condition determination result.

Accordingly, the neural network can learn the neural network 134 by labeling the training image data with external elements that are visible to the naked eye according to the characteristics of the skin.

Specifically, the external factor can be classified into at least one factor among skin tone, oiliness or dryness, pores, elasticity, and wrinkle-related information, and the neural network performs supervised learning using the labeling information for the training image. .

Accordingly, the learned neural network 134 may determine the degree of the external element from the face image of the user 10 and output the result as a predicted probability value.

Furthermore, in the present embodiment, the neural network 134 may perform the same process using the image of the input model other than the face image of the user 10 .

Referring to FIG. 6 , the neural network 134 may output a prediction probability for an external element between the first image data 304 of the user 10 and the second image data 306 of the model.

Therefore, as a result of determining the first skin condition of the output first image data 304 , it is determined the difference between the classification value of each external element and the value of the external element as the result of determining the second skin condition of the second image data 306 . And the user 10 can determine the functional elements of the cosmetics required to be similar to the image of the model.

Specifically, the cosmetic recommendation unit 130 may verify the recommended cosmetic information according to the cosmetic ingredient information related to the difference in the external elements.

That is, it is checked whether or not it contains an ingredient that can satisfy the improvement required for the image requested by the user 10 by using the corresponding ingredient information or function information among the cosmetics included in the recommended cosmetic information, and, accordingly, finally It is also possible to decide which cosmetics to recommend.

In addition, as a result according to FIG. 5 , as a result of a convolution operation on the image data 304 , it is also possible to output an N-dimensional feature map itself in a form in which the features of the external elements are emphasized.

Accordingly, in the case of FIG. 6 , as a result of the determination of the first skin condition of the outputted first image data 304 , the first feature map in which the external element is emphasized and the second skin condition of the second image data 306 are extracted as a result of determination. It is also possible to determine the difference between the second feature maps and calculate the difference for each external element from the result according to the distance between the feature vectors included in both feature maps or the cosine similarity. By using the above calculated difference, the functional elements of cosmetics necessary for the user 10 to become similar to the image of the model can be extracted.

The above cosmetic recommendation system 100 according to the present invention receives an additional input of an image that the user 10 wants to pursue, and compares the image to be pursued with the state of the wise user 10 and recommends cosmetics that are helpful for improvement. can provide

Also, the cosmetic recommendation system 100 according to the present embodiment may additionally refer to recommendation information using personal histories for each user 10 .

For example, a change in the user 10's question and answer may be used.

That is, when a changed response to the same question occurs among a plurality of questions and answers, at this time, cosmetics recommended corresponding to the previous response may be excluded from the recommendation.

Specifically, cosmetics that are recommended or excluded due to the influence of the changed response items in the recommendation table 132 may be extracted and excluded from the recommendation.

Also, the cosmetic recommendation system 100 may refer to the cosmetic purchase history of the actual user 10 .

That is, it further includes a purchase history collecting unit 140 , and the purchase history collecting unit 140 may receive the actual purchase information of the user 10 from the online platform 200 and reflect it in the recommendation.

For example, when the user 10 purchases the actually recommended cosmetics and the subsequent question and answer changes in a negative way, the cosmetics recommendation may be excluded.

Alternatively, if positive matters are derived as a result of judging the image data of the past cosmetic purchase and the external elements of the current image data using the output of the neural network, the corresponding cosmetic product may be re-recommended. In the opposite case, it can be excluded.

Accordingly, it is possible to find and provide more preferred customized cosmetics by referring to the skin condition information and actual purchase history information of the user 10 from the online platform 200 for purchasing various cosmetics.

Hereinafter, a curation-based cosmetic recommendation method will be described with reference to FIG. 7 .

First, the cosmetic recommendation method collects user response information for each step in a standardized form from an online platform (S100).

Next, the user 10 receives request information in a personalized form including image data from the user 10 (S200).

When the response information and the request information are collected, the recommended cosmetic information is generated by referring to the table defining the recommended cosmetics for each user 10 response according to the user 10 response (S300), but the output of the cosmetic recommendation neural network for the user request information In consideration of the information, the recommended cosmetic information is verified (S400).

In this embodiment, in addition to the generation and verification of recommended cosmetics sequentially performed, it is also possible to generate recommendation information using a recommendation table and neural network-based recommendation information in parallel and combine them to generate one recommended cosmetic information. possible.

In this case, the neural network may be used to determine a difference in external elements between the first image data of the user 10 and the second image data of the request model.

That is, in the verifying step ( S400 ), the recommended cosmetic information may be verified or additional cosmetic recommendation information may be generated according to the cosmetic ingredient information related thereto through the difference between the images of the external elements output from the neural network.

Specifically, as described above, the neural network may be trained to classify and output elements that visually affect an image of a person from image data.

In this case, the external element may include at least one element of skin tone, oiliness or dryness, pores, elasticity, and wrinkle-related information, and the external element may be further added or excluded, and is not limited to the exemplified element. .

Furthermore, in the collecting step ( S100 ), response change information may be extracted from the response information of the user 10 collected for a plurality of inquiries, and may be further used for recommendation of cosmetics. Therefore, in the verifying step (S400), when generating recommended cosmetic information by reflecting the change information of response information, if a response occurs in a negative direction, the previously recommended cosmetic may be excluded.

In addition, the curation-based cosmetic recommendation method further includes the step (not shown) of the purchase history collecting unit 140 receiving the actual purchase information of the user 10 from the online platform, and the step of verifying (S400) is the actual purchase information. It is also possible to generate the recommended cosmetic information by reflecting the purchase information.

In addition, the cosmetic recommendation system 100 may determine the difference in actual purchase information for each user 10 according to the same recommended cosmetic information and provide it to the online platform 200 in reverse.

Through this, the online platform 200 provides the online platform with this information to the online platform if there is a difference between whether an actual purchase is made for each user 10 having the same response and whether an actual purchase is made for each user 10 having the same response. It can provide curation information.

Furthermore, various embodiments described herein may be implemented in a computer-readable recording medium using, for example, software, hardware, or a combination thereof.

According to the hardware implementation, the embodiments described herein are ASICs (application specific integrated circuits), DSPs (digital signal processors), DSPDs (digital signal processing devices), PLDs (programmable logic devices), FPGAs (field programmable gate arrays, It may be implemented using at least one of processors, controllers, micro-controllers, microprocessors, and other electrical units for performing functions. The described embodiments may be implemented in the control module itself.

According to the software implementation, embodiments such as the procedures and functions described in this specification may be implemented as separate software modules. Each of the software modules may perform one or more functions and operations described herein. The software code may be implemented as a software application written in a suitable programming language. The software code may be stored in the memory module and executed by the control module.

The above description is merely illustrative of the technical idea of the present invention, and various modifications, changes, and substitutions are possible within the range that does not depart from the essential characteristics of the present invention by those of ordinary skill in the art to which the present invention pertains. will be.

Accordingly, the embodiments disclosed in the present invention and the accompanying drawings are for explaining, not limiting, the technical spirit of the present invention, and the scope of the technical spirit of the present invention is not limited by these embodiments and the accompanying drawings . The protection scope of the present invention should be interpreted by the following claims, and all technical ideas within the scope equivalent thereto should be construed as being included in the scope of the present invention.

Claims (12)

In the curation-based cosmetic recommendation system,
a response information collecting unit that collects user response information for each step of a question in a standardized form from the online platform;
a request information receiving unit for receiving user request information in a personalized form including image data from the user; and
A cosmetic recommendation unit for generating recommended cosmetic information with reference to a table defining recommended cosmetics for each user response according to the user response, and verifying the recommended cosmetic information in consideration of output information of a cosmetic recommendation neural network for the user request information Cosmetic recommendation system, characterized in that. The method of claim 1,
The neural network determines the difference between the user's first image data and the second image data of the request model,
The cosmetic recommendation system, characterized in that the cosmetic recommendation unit verifies the recommended cosmetic information according to the cosmetic ingredient information associated with the difference in the external factors. 3. The method of claim 2,
The external element is a cosmetic recommendation system, characterized in that it includes at least one element of skin tone, oiliness or dryness, pores, elasticity, and wrinkle-related information. The method of claim 1,
The response information collection unit extracts response change information from the user's response information collected for a plurality of queries,
The cosmetic recommendation system, characterized in that the cosmetic recommendation unit generates the recommended cosmetic information by reflecting the response change information. The method of claim 1,
The cosmetic recommendation system is
Further comprising a purchase history collecting unit for receiving the actual purchase information of the user from the online platform,
Cosmetic recommendation system, characterized in that the cosmetic recommendation unit generates the recommended cosmetic information by reflecting the actual purchase information. 6. The method of claim 5,
The purchase history collecting unit determines the difference in actual purchase information for each user according to the same recommended cosmetic information and provides the information to the online platform. In the curation-based cosmetic recommendation method,
collecting user response information for each step of a query in a standardized form from an online platform;
receiving user request information in a personalized form including image data from the user; and
Generating recommended cosmetic information with reference to a table defining recommended cosmetics for each user response according to the user response, and verifying the recommended cosmetic information in consideration of output information of a cosmetic recommendation neural network for the user request information Cosmetic recommendation method, characterized in that. 8. The method of claim 7,
The neural network determines the difference between the user's first image data and the second image data of the request model,
The verifying step is a cosmetic recommendation method, characterized in that the verification of the recommended cosmetic information according to the cosmetic ingredient information associated with the difference in the external factors. 9. The method of claim 8,
The cosmetic recommendation method, characterized in that the external factor includes at least one of skin tone, oiliness or dryness, pores, elasticity, and wrinkle-related information. 8. The method of claim 7,
The collecting step extracts response change information from the user's response information collected for a plurality of queries,
The verifying step reflects the response change information to generate the recommended cosmetic information. 8. The method of claim 7,
The cosmetic recommendation method is,
Further comprising a purchase history collecting unit for receiving the actual purchase information of the user from the online platform,
The verifying step reflects the actual purchase information to generate the recommended cosmetic information. 12. The method of claim 11,
The purchase history collection unit determines the difference in actual purchase information for each user according to the same recommended cosmetic information, and provides the information to the online platform.

Images