[go: up one dir, main page]

WO2021071238A1 - Procédé de recommandation de produit de mode, dispositif et programme informatique - Google Patents

Procédé de recommandation de produit de mode, dispositif et programme informatique Download PDF

Info

Publication number
WO2021071238A1
WO2021071238A1 PCT/KR2020/013648 KR2020013648W WO2021071238A1 WO 2021071238 A1 WO2021071238 A1 WO 2021071238A1 KR 2020013648 W KR2020013648 W KR 2020013648W WO 2021071238 A1 WO2021071238 A1 WO 2021071238A1
Authority
WO
WIPO (PCT)
Prior art keywords
user
image
preference
item
information
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/KR2020/013648
Other languages
English (en)
Korean (ko)
Inventor
유애리
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Odd Concepts Inc
Original Assignee
Odd Concepts Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Odd Concepts Inc filed Critical Odd Concepts Inc
Publication of WO2021071238A1 publication Critical patent/WO2021071238A1/fr
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • G06Q30/0601—Electronic shopping [e-shopping]
    • G06Q30/0631—Recommending goods or services
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0278—Product appraisal
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • G06Q30/0601—Electronic shopping [e-shopping]
    • G06Q30/0641—Electronic shopping [e-shopping] utilising user interfaces specially adapted for shopping
    • G06Q30/0643—Electronic shopping [e-shopping] utilising user interfaces specially adapted for shopping graphically representing goods, e.g. 3D product representation

Definitions

  • the present invention relates to a method of recommending a fashion product. More specifically, the present invention relates to a fashion product recommendation system that generates user preference information based on a user selection image reflecting the user preference and provides a recommended item to a user based on the user preference information.
  • An object of the present invention is to provide a method, an apparatus, and a computer program for recommending fashion products having improved search capability.
  • a service server that generates user preference information, which is information that can estimate a user's preference for a specific fashion item, an image of a single fashion item or a style image in which a plurality of fashion items are well suited.
  • a diagnostic image generator that includes a preference diagnosis item reflecting the user's taste and generates a diagnostic image for identifying the user's preference, and provides the diagnostic image to the user device, an image selected by the user among the diagnostic images
  • An update information generation unit receiving the user-selected image from the user device and generating the user preference information from fashion items included in the user-selected image or the fashion item purchased by the user, and a user storing the user preference information It characterized in that it comprises a preference information storage unit.
  • a method, apparatus, and computer program for recommending fashion products having improved search capability can be provided.
  • FIG. 1 is a diagram illustrating a fashion product recommendation system according to an exemplary embodiment of the present invention.
  • FIG. 2 is an apparatus diagram for explaining the operation of the fashion product recommendation system of FIG. 1.
  • FIG. 3 is a diagram for explaining a process of generating a diagnostic image from a preference diagnostic item image according to an embodiment of the present invention.
  • FIG. 4 is a diagram illustrating a process of generating a user-selected image according to an embodiment of the present invention.
  • FIG. 5 is a flowchart illustrating generation of user preference information according to an embodiment of the present invention.
  • FIG. 6 is a flowchart illustrating generation of user preference information according to another embodiment of the present invention.
  • first and/or second may be used to describe various components, but the components should not be limited by the terms. The terms are only for the purpose of distinguishing one component from other components, for example, without departing from the scope of the rights according to the concept of the present invention, the first component may be named as the second component, and similarly The second component may also be referred to as a first component.
  • FIG. 1 is a diagram illustrating a fashion product recommendation system according to an exemplary embodiment of the present invention.
  • the fashion product recommendation system 50 may include a user device 100 and a service server 200.
  • the user device 100 may be a concept including all types of electronic devices capable of requesting search and displaying advertisement information, such as a desktop, a smartphone, and a tablet PC.
  • the service server 200 may include a user preference information generation unit 210 and a user preference information storage unit 220.
  • the user preference information generation unit 210 may generate user preference information, which is information capable of estimating a user's preference for a specific fashion item such as design, price range, brand, and use.
  • the user preference information may include vector values for calculating a user preference for a specific fashion item.
  • the preference information generation unit 210 may collect user behavior information for a fashion item in order to generate user preference information. For example, fashion item information browsed using a user device, fashion item information shared through a social network service or a messenger service, and fashion item information stored or purchased in a shopping cart of an online shopping mall may be collected. Furthermore, it is also possible to collect images of pre-owned fashion items generated by the user device using the camera module.
  • the preference information generation unit 210 may collect the fashion item information in the form of an image of the fashion item or a fashion item object image included in an image. At this time, the fashion item object image may be stored together with the arbitrary image, and the coordinate value of the object area for the arbitrary image, the image in which the object area is cropped in the arbitrary image, and/or the cropped image. It may include at least one or more of a category, a property, and a vector value. According to an embodiment of the present invention, the preference information generation unit 210 may generate a query for determining the user's preference. For example, when there is no behavioral information for a certain user, preference information of a corresponding user may be generated by using user response information for the query. As another example, preference information of a corresponding user may be more accurately generated by using user response information to the query together with user behavior information.
  • the preference information generator 210 may generate a diagnostic image including at least one preference diagnostic item as a query and provide it to the user device 100.
  • the diagnostic image may include fashion items having one or more different attributes within the same category.
  • the preference information generation unit 210 in order to diagnose the user's preference for the cardigan category, the preference information generation unit 210 generates a cardigan image having at least one or more different attributes such as color, pattern, neckline, material, and sleeve length as a diagnosis image.
  • User preference information may be generated by using the user's selection information.
  • the user preference information generation unit 210 may generate the diagnosis image using a lookbook database.
  • the lookbook may include an image of a single fashion item or an image (hereinafter referred to as a style image) in which a plurality of fashion items are combined to suitably fit among images collected online.
  • Style images include wearing shots of products sold in the online market, fashion show shooting images, fashion magazine pictorial images, SNS, blog celebrity costume images, idol costume images, fashion magazine street fashion images or fashion items For example, the image coordinated with other items for the sale of the product can be illustrated.
  • the preference diagnosis item may be a fashion item in which a user's taste is reflected.
  • the preference diagnosis item may be a fashion item estimated to be preferred by the user based on an item previously purchased by the user or an item held by the user.
  • the preference diagnosis item may be a fashion item in which the user responds that his or her taste is reflected at the request of the service server 200.
  • the service server 200 requests a user to take a picture of an image of a preference diagnosis item, requests input of a keyword to describe the preference diagnosis item, requests a voice command through a microphone, etc.
  • Information on a preference diagnosis item may be collected through the user preference information generator 210.
  • the user preference information generation unit 210 may receive information on a user selection image, which is a diagnosis image selected by the user, from the user device 100.
  • the user can determine a favorite image among the diagnostic images as the user-selected image.
  • a user may abstractly know what his or her taste is, but in order to search for a fashion item that specifically reflects his or her taste, there has been a cumbersome need to search including all keywords describing the taste.
  • a user simply selects a single fashion item provided by the service server 200 or a favorite image from among a plurality of style images (diagnostic images) to set his or her taste to the service server 200. ), and thus the efficiency of the search can be increased.
  • the user preference information generation unit 210 may generate user preference information using the user selection image.
  • the user preference information generation unit 210 may generate a user preference information vector by combining feature information of fashion items included in the user selection image.
  • the user preference information generation unit 210 may generate user preference information from fashion items purchased by the user in addition to the user selection image.
  • the fashion item purchased by the user may be an item reflecting the user's preference. Accordingly, the user preference information generation unit 210 may generate user preference information by combining feature information included in fashion items purchased by the user.
  • the service server determines that the information received from the user is insufficient, and the user device 100 receives a preference diagnosis item in order to extract the diagnostic image again. You can request an image of.
  • the service server 200 determines that the diagnostic image does not sufficiently reflect the user's preference information, and may directly receive an image of the preference diagnostic item through the user in order to regenerate the diagnostic image.
  • the service server 200 may perform a preference survey again when the user purchases a preference diagnosis item or a product different from the recommended item, which is data for confirming the user's preference.
  • the service server 200 may generate a diagnosis image based on a preference diagnosis item, and update user preference information based on the diagnosis image selected by the user.
  • the service server 200 provides a recommended item determined based on user preference information to a user, and the user may purchase a fashion item different from the recommended item or an item different from a preference diagnosis item provided as a recommendation item.
  • the service server 200 may determine that the previously performed preference diagnosis is wrong and perform a preference survey again.
  • the process of extracting the preference diagnosis item from the image of the conventionally purchased item or the possessed item described above may be repeated, or a diagnosis image may be generated again by receiving a preference diagnosis item directly from the user.
  • the service server 200 may provide a message indicating that the diagnostic image did not sufficiently reflect the user's preference, a message indicating that it is difficult to provide an appropriate recommended item due to insufficient number of user-selected images, or a user who provided precautions.
  • a message for requesting input of information on a preference diagnosis item together with a message notifying that a new diagnosis image may be provided may be transmitted to the user device 100.
  • the service server 200 diagnoses preferences through various forms, such as requesting a user to take a picture of an image of a preference diagnosis item, requesting input of a keyword that can describe a preference diagnosis item, or requesting a voice command through a microphone. You can send a message requesting information on an item.
  • the user preference information generation unit 210 may estimate the user's preference by using the user's behavior information for the fashion item and/or the user's response information to the preference diagnosis query.
  • the user preference information generation unit 210 may extract information on a category or attribute from an image selected by the user from among diagnostic images provided as a query for user preference survey, and generate an initial user preference vector using this. . Furthermore, the user preference information generation unit 210 acquires an image of an arbitrary fashion item purchased by the user as a preference image, extracts information on the category or attribute of the preference image, and reflects the extracted information in the user preference vector. User preference information can be updated in a way.
  • the service server 200 may store a product image or a style image in the form of a vector value. Specifically, the service server 200 may detect feature areas of product images or style images (Interest Point Detection). The feature region may mean a descriptor for a feature of an image for determining whether or not the images are identical or similar, that is, a main region for extracting a feature description.
  • such a characteristic area is a contour line included in an image, a corner such as a corner among the contours, a blob distinguished from the surrounding area, an area that is invariant or covariable according to the transformation of the image, or more They can be poles with dark or bright features, and can target patches (fragments) of the image or the entire image.
  • the service server may extract a feature descriptor from the feature domain.
  • the feature descriptor may be a representation of the features of an image as a vector value.
  • such a feature descriptor may be calculated by using the position of the feature region with respect to the corresponding image, or brightness, color, sharpness, gradient, scale, or pattern information of the feature region.
  • the feature descriptor may be calculated by converting a brightness value, a brightness change value, or a distribution value of the feature region into a vector.
  • a feature descriptor for an image is not only a local descriptor based on a feature region as described above, but also a global descriptor, a frequency descriptor, a binary descriptor, or It can be expressed as a neural network descriptor.
  • the feature descriptor is a global descriptor ( Global descriptor).
  • a feature descriptor is a frequency descriptor that converts and extracts the number of times that specific descriptors classified in advance are included in an image and the number of global features such as a conventionally defined color table, etc.
  • Learning at the layer of a binary descriptor or neural network that extracts bitwise whether or not the values of each element constituting the descriptor are larger or smaller than a specific value and converts it to an integer type.
  • it may include a neural network descriptor that extracts image information used for classification.
  • Machine learning is one of the fields of artificial intelligence, and can be defined as a set of systems and algorithms for learning based on empirical data, performing predictions, and improving its own performance.
  • the models used by the service server include Deep Neural Networks (DNN), Convolutional Deep Neural Networks (CNN), Reccurent Neural Networks (RNN), and deep trust neural networks among these machine learning models. Deep Belief Networks, DBN) may be used.
  • a feature information vector extracted from a product image or a style image may be converted to a lower dimension.
  • feature information extracted through an artificial neural network corresponds to 40,000-dimensional high-dimensional vector information, and it may be appropriate to convert it into a low-dimensional vector in an appropriate range in consideration of resources required for search.
  • the feature information vector may be converted using an algorithm such as PCA, and the feature information converted into a low-dimensional vector may be indexed into a corresponding product image.
  • vector values included in the user-selected image may include brightness, color, sharpness, gradient, scale, or pattern information of the user-selected image.
  • the user preference information generation unit 210 may generate user preference information from a combination of these information.
  • user preference information may be generated using label information extracted from fashion item objects included in a user selection image. For example, when a label for describing the content of a corresponding fashion item is previously set in a fashion item included in a user selection image, user preference information may be generated by combining labels set in a plurality of selection images.
  • user preference information may be generated using information on a preference diagnosis item selected by the user as well as fashion items previously purchased by the user and/or fashion items previously held by the user. have.
  • a weight for combining information on the preference diagnosis item selected by the user, the purchased item, and/or the items possessed may be set according to various algorithms.
  • user preference information may be continuously updated by receiving additional preference surveys such as'good' or'notice' from the user.
  • On/offline markets can continuously release new products that reflect the changing needs of consumers.
  • New products released may contain designs (or label information) that existing products do not cover.
  • the existing design (or label information) can be reinterpreted and re-created as a new combination of design (or label information) that stimulates the consumer's purchase desire that was not in the existing online/offline market.
  • These new products may reflect the latest needs of consumers the most, so user preference information needs to be updated.
  • the service server 200 may update user preference information through a query such as'good or not' about the new product. Depending on the user's response, preferences may be reflected in user preference information with different weights.
  • the user preference information may include information capable of estimating a user's preference, such as a design, a brand, and a price point of a fashion item.
  • the user preference information may include information on a brand of a fashion item that the user prefers.
  • User preference information on the brand may be generated from vector values included in the user-selected image.
  • the service server 200 may recognize a feature part (eg, a tag attached to clothing, an inner part of the collar of a top, a chest part, an upper part of the bottom, a pocket part, etc.) in which the brand is displayed in the user-selected image, User preference information about the brand can be generated by searching the database for characteristic information of the corresponding brand.
  • characteristic information of individual brands such as logos, check patterns, figures, patterns, trademarks, and phrases, may be stored in advance.
  • the service server 200 may extract characteristic information of a brand from a vector value of a user-selected image, and generate user preference information about a brand that the user prefers.
  • user preference information may be generated from label information extracted from fashion item objects included in a user selection image.
  • Label information about a brand of a corresponding fashion item may be tagged in advance in fashion item objects included in the user-selected image.
  • the service server 200 may generate user preference information about the brand by combining label information about the brand.
  • the user preference information storage unit 220 may store user preference information received from the user preference information generation unit 210.
  • the usage preference information stored in the user preference information storage unit 220 may be updated each time a user selection image is input.
  • User preference information may change over time. For example, as the user's personal information or surrounding environment changes, such as the user's age group, financial strength, job, and workplace atmosphere, the preferred style may change. In addition, user preference information reflecting the currently popular style may be newly generated according to the change of the latest trend.
  • a number of fashion items giving a casual feel such as a colored T-shirt, jeans, sneakers, or sneakers
  • a number of fashion items giving an office look such as white shirts, shoes, neckties, and achromatic fashion items, may be included in the user preference information.
  • the user preference storage unit 220 may aggregate user preference information for a single fashion item or lookbook for each user, and arrange and provide the user preference information for each user. To this end, user preference information may be tagged with a user ID assigned to each user and stored together. The user selects the diagnosis image received from the service server 200 from time to time, at an arbitrary timing, or periodically, thereby selecting a changed taste. Can be delivered to the service server. As the user preference information is updated each time a user-selected image is input, the service server 200 can more accurately predict a style desired by the user and recommend a fashion item corresponding thereto.
  • FIG. 2 is an apparatus diagram for explaining the operation of the fashion product recommendation system of FIG. 1.
  • the service server 200 may include a user preference information generation unit 210 and a user preference information storage unit 220.
  • the service server 200 receives from the user device 100 a conventionally purchased item, an image of a possessed item, etc. that can estimate the user's taste, and diagnoses to determine the user's preference for a specific fashion item based on this You can provide an image.
  • the service server 200 may generate user preference information by calculating vector values included in the user-selected image.
  • the user preference information generation unit 210 may include an initial user preference information generation unit 211, a diagnostic image generation unit 212, and an update information generation unit 213.
  • the initial user preference information generation unit 211 may receive an image of an item previously purchased by the user or an item held by the user from the user device 100.
  • the image of an item previously purchased by a user or an item held by the user may be requested from the user device 100 when the service server 200 requires it to update user preference information.
  • a user purchases a fashion item in an online market it may be provided to the service server 200 at an arbitrary time point or at a certain period.
  • an image of an item previously purchased or an item held by a user is exemplified as information for generating initial user preference information, but according to an embodiment, the initial user preference information generation unit 211 is a blog browsing time for a specific fashion item. , The number of visits, text, a video, or a URL of a web site on which a preference diagnosis item is displayed, etc. may be included.
  • the initial user preference information generation unit 211 may generate initial user preference information from an image of a conventionally purchased item or possessed item.
  • the initial user preference information may be information on a user's preference extracted from an image of an item previously purchased by the user or an item held by the user.
  • the initial user preference information may include vector values for calculating a user preference for a specific fashion item.
  • Initial user preference information may be extracted based on various information such as blog browsing time, number of visits, text, etc. for a specific fashion item.
  • the initial user preference information may include characteristic information of an item previously held by the user. That is, the initial user preference information may include information on a user's taste extracted from an item previously held by the user.
  • the initial user preference information is based on past data that the user has previously purchased or already possessed items, the user's changed taste may not be reflected.
  • a preference diagnosis item may be extracted based on initial user preference information, and user preference information reflecting a changed user's taste may be generated through at least one diagnosis image including the extracted preference diagnosis item.
  • the diagnostic image generator 212 may generate a diagnostic image from initial user preference information and provide it to the user device 100.
  • the diagnostic image may be an image including at least one preference diagnostic item retrieved from a lookbook in order to grasp the user's preference. Generation of the diagnostic image will be described later in the description of FIG. 3.
  • the user device 100 may select a favorite image from among the diagnostic images received from the diagnostic image generator 212.
  • the diagnostic image selected by the user may be a user-selected image.
  • the update information generator 213 may generate user preference information from a fashion item object included in the user selection image.
  • the user preference information may be information capable of estimating a user's preference for a specific fashion item, such as a design, a price point, and a brand.
  • the user preference information may include vector values for calculating a user preference for a specific fashion item.
  • the service server 200 may convert these features into vector values and store them as user preference information.
  • the update information generation unit 213 may generate user preference information from fashion items purchased by the user in addition to the user selection image.
  • the fashion item purchased by the user may be an item reflecting the user's preference. Accordingly, the update information generation unit 213 may generate user preference information by combining feature information included in fashion items purchased by the user.
  • the user preference information may be extracted by calculating vector values included in fashion items of the user-selected image.
  • user preference information may be extracted through a calculation process in which more vector values are substituted.
  • the more the calculation process is performed the more sophisticated the user preference information, and the user preference information can reflect the user's preference in more detail and accurately.
  • weights may be assigned and priorities may be determined according to the number of overlaps. For example, if the “cross bag” is duplicated 4 times and the “skirt” is duplicated twice, it may be determined that the user's preference for the “cross bag” is higher than that of the “skirt”.
  • the service server 200 may assign a weight of higher user preference to these features.
  • the feature in which the weight of the higher user preference is reflected may be prioritized in the database when determining the recommended item.
  • the service server may request an image of the preference diagnosis item from the user device 100 in order to extract the diagnosis image again.
  • the service server 200 determines that the diagnostic image does not sufficiently reflect the user's preference information, and may directly receive an image of the preference diagnostic item through the user in order to regenerate the diagnostic image.
  • the service server may perform a preference survey again when a user purchases a product different from a preference diagnosis item, which is data for confirming the user's preference.
  • the service server may generate a diagnosis image based on the preference diagnosis item, and update user preference information based on the diagnosis image selected by the user.
  • the service server provides the user with a recommended item determined based on user preference information, and the user may purchase a fashion item different from the recommended item or an item different from the preference diagnosis item provided as a recommendation item.
  • the service server may determine that the previously conducted preference diagnosis is wrong and perform a preference survey again.
  • the process of extracting the preference diagnosis item from the image of the previously purchased item or the possessed item is repeated to regenerate a new preference diagnosis item, or a preference diagnosis item is directly input from the user and diagnosed. You can recreate the image.
  • the service server 200 may provide a message indicating that the diagnostic image did not sufficiently reflect the user's preference, a message indicating that it is difficult to provide an appropriate recommended item due to insufficient number of user-selected images, or a user-selected image provided as a precaution.
  • a message requesting input of information on a preference diagnosis item together with a message notifying that a new diagnosis image may be provided may be transmitted to the user device 100.
  • the service server 200 diagnoses preferences through various forms, such as requesting a user to take a picture of an image of a preference diagnosis item, requesting input of a keyword that can describe a preference diagnosis item, or requesting a voice command through a microphone. A message requesting information on an item can be transmitted.
  • the update information generation unit 213 may store the generated user preference information in the user preference information storage unit 220.
  • Existing user preference information stored in the user preference storage unit 220 may be updated by reflecting newly received user preference information.
  • the user preference information may be continuously updated by receiving additional preference surveys such as'good' or'notice' from the user.
  • On/offline markets can continuously release new products that reflect the changing needs of consumers.
  • New products released may contain designs (or label information) that existing products do not cover.
  • existing designs (or label information) can be reinterpreted and re-created as a new combination of designs (or label information) that stimulates consumers' desire to purchase that did not exist in the existing online/offline market. .
  • These new products may reflect the latest needs of consumers the most, so user preference information needs to be updated.
  • the service server 200 may update user preference information through queries such as'good,'normal', and'by' about the new product. Depending on the user's response, preferences may be reflected in user preference information with different weights.
  • FIG. 3 is a diagram for explaining a process of generating a diagnostic image from a preference diagnostic item image according to an embodiment of the present invention.
  • FIG. 1 may be an image of a preference diagnosis item that is determined to reflect a user's taste from an image of an item purchased or possessed by a user.
  • the user may be an image of a preference diagnosis item input from the user.
  • the service server may generate at least one diagnostic image including an image of the preference diagnostic item.
  • ten diagnostic images may be generated, but fewer or more than ten diagnostic images may be generated.
  • the diagnosis image may be an image obtained by searching a lookbook including a plurality of style images for a style image coordinated with a preference diagnosis item. If there are a number of style images coordinated with the preference diagnosis item in the lookbook, the service server may remove the overlapping style image. For example, not only the completely identical style image, but also the coordinated fashion items may be the same, but only the poses of the person wearing the fashion item or the mannequin may be different. In this case, the service server may determine only style images having different sets of fashion items to be included as diagnostic images.
  • the number of style images including the preference diagnosis item in the lookbook may be smaller than a preset value or may not exist.
  • the service server may determine only the searched style image as a diagnosis image, may determine another style image matched with a fashion item similar to the preference diagnosis item as the diagnosis image, or notify the user device that the diagnosis image has not been searched.
  • a message requesting input of information on the preference diagnosis item may be transmitted to the user device.
  • Fig. 1a, Fig. 1b and Fig. 1c show that "the alphabet is written on a white background in the middle part of the clothes, and the upper part is red and the lower part is blue.”
  • a user device in which three diagnostic images (diagnostic image 1, diagnosis image 2, and diagnosis image 3) generated from the image of the preference diagnosis item "" are displayed.
  • the diagnostic image 1, the diagnostic image 2, and the diagnostic image 3 may be different style images coordinated with the same preference diagnostic item.
  • diagnostic image 1 is a blue skirt and a red handbag
  • diagnostic image 2 is a T-shirt and white leggings with an alphabet in the center of the clothing
  • diagnostic image 3 is a T-shirt and black cotton pants with an alphabet on the left side of the chest. It is coordinated with diagnostic items.
  • the service server may sequentially provide the generated diagnostic images to the user device.
  • Figure 1a shows the diagnostic image 1 displayed on the user device. If it is determined that the blue skirt and the red handbag in the diagnostic image 1 are their favorite fashion items, the user can click the heart icons 40a, 40b, and 40c indicating "like". On the contrary, if it is determined that the blue skirt is not a fashion item that he or she prefers, the X icons 50a, 50b, 50c, which mean "I don't like,” may be clicked.
  • the user can also reflect his or her preference for the diagnostic image displayed on his or her screen by dragging the diagnostic image or the user interface screen in a preset direction. Drag left, “dislike” drag right)
  • preferences for the diagnostic image may be transmitted to the service server in various ways, such as clicking or double-clicking a specific area of the diagnostic image, sending a voice command to a microphone, or clicking a separate button set in advance.
  • User-selected image display windows are located at the bottom of the diagnostic image (diagnostic image 1, diagnostic image 2, diagnostic image 3), heart icons (40a, 40b, 40c) and X icons (50a, 50b, 50c). Can be located.
  • the user-selected image display windows 60a, 60b, and 60c may be a specific area of a user interface in which a user-selected image, which is a diagnostic image selected as "Like" by the user, is displayed as a thumbnail.
  • a user-selected image which is a diagnostic image selected as "Like” by the user
  • a process in which the user-selected image is displayed as a thumbnail will be described later in the accompanying description of FIG. 4.
  • FIG. 4 is a diagram illustrating a process of generating a user-selected image according to an embodiment of the present invention.
  • FIG. 4A is a user interface that asks whether or not the user prefers diagnosis image 1, and when the user selects “Like” for diagnosis image 1, the diagnosis is displayed on the user-selected image display window 60d.
  • Fig. 2b is a user interface in which image 1 is displayed as a thumbnail.
  • the user may click the heart icon 40d or the X icon 50d.
  • the user may click the heart icon 40d when it is determined that the diagnosis image 1 reflects his or her taste, and click the X icon 50d when it is determined that the diagnosis image 1 does not reflect his or her taste.
  • the diagnostic image is displayed on the user-selected image display window (60d) as shown in Figure 2b, and the next diagnostic image (diagnosis image 2) is sequentially displayed on the user interface.
  • the diagnostic image displayed on the user-selected image display window 60d that is, the diagnostic image selected by the user may be a user-selected image.
  • the diagnostic image is not displayed on the user-selected image display window 60d, and the next diagnostic image (diagnosis image 2) can be sequentially displayed on the user interface. .
  • FIG. 4B is a diagram showing a user interface asking whether or not the user prefers a diagnosis image 2, and when the user selects “Like” for the diagnosis image 2, the diagnosis image 2 is displayed on the user-selected image display window.
  • Fig. 2d is a user interface displayed as a thumbnail.
  • the user may select a heart icon 40d indicating “Like”.
  • the diagnostic image 2 is displayed on the user-selected image display window, and although not shown in the drawing, it is possible to sequentially ask the user whether or not to prefer the next diagnostic image (diagnosis image 3).
  • the diagnostic image 2 is not displayed on the user-selected image display window, and the next diagnostic image (diagnostic image 3) is sequentially displayed on the user interface. Can be displayed.
  • the user-selected image display window may include a number of user-selected images according to the user's “Like” selection.
  • FIG. 4A shows that one user-selected image (user-selected image 1) is displayed as a thumbnail, as shown in FIG. 4B, a plurality of user-selected images may be displayed on the user-selected image display window 60d.
  • the user preference information may be extracted by calculating vector values included in fashion items of the user-selected image. Accordingly, as the number of user-selected images increases, user preference information may be extracted through a calculation process in which more vector values are substituted. The more the calculation process is performed, the more sophisticated the user preference information, and the user preference information can reflect the user's taste in more detail and accurately.
  • the service server may extract user preference information based on only the user-selected image 1.
  • fashion items that match the preference diagnosis item may be a blue skirt and a red handbag.
  • the service server may generate user preference information by combining vector values including brightness, color, sharpness, gradient, scale, or pattern information of the image of the blue skirt and red handbag.
  • the user preference information may include information capable of estimating a user's preference, such as a design, price point, brand, and use of a fashion item.
  • the service server obtains a vector value corresponding to “blue” and “skirt” from an image of a blue skirt, and a vector value corresponding to “red”, “cross bag”, and “quilt pattern” from an image of a red handbag, respectively. Can be extracted.
  • the service server combines these vector values, and it can be estimated that the user prefers fashion items having the characteristics of “blue”, “skirt”, “red”, “cross back”, and “quilt pattern”.
  • weights may be assigned and priority may be determined according to the number of overlaps. For example, if the “cross bag” is duplicated 4 times and the “skirt” is duplicated twice, it may be determined that the user's preference for the “cross bag” is higher than that of the “skirt”. This information may be user preference information stored in the form of a vector value.
  • Figure 2d shows a user interface in which a plurality of user-selected images are selected (user-selected image 1 and user-selected image 2 are selected).
  • the service server extracts user preference information based on the user-selected image 1 and the user-selected image 2 can do.
  • the service server retrieves user preference information indicating that the user prefers fashion items having the features of “blue”, “skirt”, “red”, “cross back”, and “quilt pattern”. Can be generated.
  • fashion items matched with the preference diagnosis item may be a t-shirt and white leggings with an alphabet written in the center of the clothes.
  • the service server may generate user preference information by combining vector values including brightness, color, sharpness, gradient, scale, or pattern information of the T-shirt and white leggings image with the alphabet in the center of the clothes.
  • the service server obtains a vector value corresponding to “alphabet print”, “white T-shirt”, and “pattern in the center of clothes” from the image of the T-shirt in which the alphabet is written in the center of the clothes, and the image of the white leggings.
  • Vector values corresponding to "white” and “leggings” can be extracted from from.
  • the service server combines these vector values, and the user prefers fashion items with features such as “alphabet print”, “white t-shirt”, “pattern in the center of the clothes”, “white”, and “leggings”. It can be assumed that it is.
  • This information may be user preference information stored in the form of a vector value.
  • the service server from user-selected image 1 and user-selected image 2 is selected from “blue”, “skirt”, “red”, “cross back”, “quilt pattern”, “alphabet print”, “white T-shirt”, and “clothes”.
  • the user preference information stored in the form of a vector value can be generated in which the characteristics of fashion items such as "patterned”, “white”, and “leggings" are stored in the center of the.
  • FIG. 5 is a flowchart illustrating generation of user preference information according to an embodiment of the present invention.
  • the service server may determine whether information on an item previously purchased by a user or an item to be retained exists.
  • the service server may perform step S503 when it is determined that information on the item previously purchased by the user or the item to be held exists, and may perform step S507 when it is determined that it does not exist.
  • the service server may generate initial user preference information from an image of an item that the user has previously purchased or possessed.
  • the initial user preference information may include characteristic information of an item previously held by the user. That is, the initial user preference information may be information on a user's preference extracted from an image of an item previously purchased by the user or an item held by the user.
  • the initial user preference information may include vector values for calculating a user preference for a specific fashion item.
  • Initial user preference information may be extracted based on various information such as blog browsing time, number of visits, text, etc. for a specific fashion item.
  • a preference diagnosis item may be extracted based on initial user preference information, and user preference information reflecting a changed user's taste may be generated through at least one diagnosis image including the extracted preference diagnosis item.
  • the service server may extract a preference diagnosis item from initial user preference information. Specifically, the service server may extract a preference diagnosis item by combining vector values included in the initial user preference information and corresponding to the characteristics of the fashion item reflecting the user's taste.
  • the preference diagnosis item may be a fashion item estimated to be preferred by a user based on an item previously purchased or possessed by the user.
  • the initial user preference information generated from information on items purchased or possessed by the user may be “feminine feeling”, “one piece”, “flower pattern”, “autumn feeling”, and “silk material”. have.
  • the initial user information may include features of the above fashion items converted into vector values that can be recognized by a computer.
  • the service server may combine these vector values and determine the fashion item closest to the vector value as a preference diagnosis item. For example, it may be determined that a vector value generated by a dot product of the above vector values corresponds to a characteristic of a specific fashion item reflecting the user's preference.
  • the service server can search for fashion items that are most similar to the generated vector values based on image similarity in a separate database, and determine the fashion items corresponding to the generated vector values and the vector values closest to the image similarity as the preference diagnosis item. I can.
  • the service server extracts image feature information of a fashion item included in a style image, and expresses the feature information as a vector value to generate a feature value of the fashion item and structure feature information of the images.
  • the service server defines the feature information of the fashion item, generates a neural network model that learns the feature of the image corresponding to the feature region, classifies the feature information in the image of the fashion item, and Feature information about the image of a fashion item can be extracted.
  • the service server may assign the corresponding feature information to an image that matches a specific pattern with a random probability through a neural network model that has learned a pattern of an image corresponding to each feature information.
  • the service server may generate feature information of a fashion item using machine learning based on a recurrent neural network (RNN).
  • Machine learning is one of the fields of artificial intelligence, and can be defined as a set of systems and algorithms for learning based on empirical data, performing predictions, and improving its own performance.
  • the models used by the service server include Deep Neural Networks (DNN), Convolutional Deep Neural Networks (CNN), Reccurent Neural Networks (RNN), and deep trust neural networks among these machine learning models. Deep Belief Networks, DBN) may be used.
  • the service server forms an initial neural network model by learning the characteristics of an image corresponding to each feature information, and applies a large number of fashion item images to it to more elaborately extend the neural network model. May be.
  • the service server may apply fashion item images to a neural network model formed in a hierarchical structure formed of a plurality of layers without additional learning about feature information. Furthermore, the feature information of the fashion item image is weighted according to the request of the corresponding layer, the fashion item images are clustered using the processed feature information, and the clustered image group is given a “feminine feeling”, “one piece”, and “flowers”. Characteristic information that is interpreted posteriorly, such as pattern”, “autumn feeling”, and “silk material”, can be assigned.
  • the service server may perform step S507.
  • the preference diagnosis item may be a fashion item in which the user responds as a fashion item reflecting his or her taste according to a query requesting the service server to select a preference diagnosis item.
  • the service server requests the user to take a picture of an image of a preference diagnosis item, requests input of a keyword that can describe the preference diagnosis item, or requests a voice command through a microphone.
  • Information on diagnostic items can be requested from the user.
  • the service server may generate a diagnostic image based on the lookbook and provide it to the user.
  • the diagnostic image may be an image including at least one preference diagnostic item retrieved from a lookbook in order to grasp the user's preference.
  • the number of style images including the preference diagnosis item in the lookbook may be smaller than a preset value or may not exist.
  • the service server may determine only the searched style image as a diagnosis image, may determine another style image matched with a fashion item similar to the preference diagnosis item as the diagnosis image, or notify the user device that the diagnosis image has not been searched.
  • a message requesting input of information on the preference diagnosis item may be transmitted to the user device. Generation of the diagnostic image may be described with reference to FIG. 3.
  • the service server may provide a query for a style image reflecting the user's taste to the user device through a diagnosis image, and may generate user preference information by receiving a user response to the query.
  • the diagnostic image selected by the user may be a user-selected image.
  • the service server may receive a user selection image from the user device.
  • a user can transmit his or her taste to the service server by simply selecting a favorite image among a plurality of style images (diagnostic images) provided by the service server, and thus the efficiency of search is improved. There is an advantage that can be increased.
  • the service server may generate user preference information based on the user selection image.
  • a process of generating user preference information may be similar to a process of generating initial user preference information in step S503.
  • the user preference information may be information capable of estimating a user's preference for a specific fashion item such as design, price range, brand, and usage.
  • the user preference information may include vector values for calculating a user preference for a specific fashion item.
  • the user preference information may include information on a brand of a fashion item that the user prefers.
  • User preference information about the brand may be generated from vector values included in the user-selected image.
  • the service server can recognize the brand-marked feature part (for example, the tag attached to the clothing, the inner part of the collar on the top, the chest part, the top part of the bottom, the pocket part, etc.) from the user-selected image, and User preference information about a brand can be generated by searching for characteristic information in a database.
  • characteristic information of individual brands such as logos, check patterns, figures, patterns, trademarks, and phrases, may be stored in advance.
  • the service server may extract characteristic information of the brand from the vector value of the user-selected image and generate user preference information about the brand that the user prefers.
  • user preference information may be generated from label information extracted from fashion item objects included in a user selection image.
  • Label information about a brand of a corresponding fashion item may be tagged in advance in fashion item objects included in the user-selected image.
  • the service server may generate user preference information about the brand by combining label information about the brand.
  • the service server can convert these features into vector values and store them as user preference information.
  • the process of generating user preference information may be described with reference to FIG. 4.
  • the service server may provide a recommended item to the user device based on the generated user preference information.
  • the service server may search a product database using user preference information and provide a product list matching the user preference information as a recommended item.
  • a product matching user preference information among products including the category and/or attribute is selected from a product database.
  • a product including a selected category and/or attribute from among products matching the user preference information may be determined as a recommended item.
  • a process of determining a recommended item based on user preference information may be similar to a process of extracting a preference diagnosis item from initial user preference information in step S505.
  • user preference information is “blue”, “skirt”, “red”, “cross back”, “quilt pattern”, “alphabet print”, “white T-shirt”, “pattern in the center of clothing” , “White” or “leggings”.
  • the initial user information may include features of the above fashion items converted into vector values that can be recognized by a computer.
  • the service server may combine these vector values and determine the fashion item closest to the vector value as a recommended item. For example, it may be determined that a vector value generated by a dot product of the above vector values corresponds to a characteristic of a specific fashion item reflecting the user's preference.
  • the service server can search for fashion items most similar to the generated vector values based on image similarity in a separate database, and determine the fashion items corresponding to the generated vector values and the vector values closest to the image similarity as recommended items. have.
  • the user can reflect the changed taste on the service server by selecting the diagnostic image received from the service server from time to time or at an arbitrary timing.
  • the service server has the effect of more accurately predicting the style desired by the user and recommending a fashion item corresponding thereto.
  • the user preference information may be continuously updated by receiving additional preference surveys such as'good' or'notice' from the user.
  • On/offline markets can continuously release new products that reflect the changing needs of consumers.
  • New products released may contain designs (or label information) that existing products do not cover.
  • existing designs (or label information) can be reinterpreted and re-created as a new combination of designs (or label information) that stimulates consumers' desire to purchase that did not exist in the existing online/offline market. .
  • These new products may reflect the latest needs of consumers the most, so user preference information needs to be updated.
  • the service server may update user preference information through queries such as'good,'normal', and'by' about the new product. Depending on the user's response, preferences may be reflected in user preference information with different weights.
  • FIG. 6 is a flowchart illustrating generation of user preference information according to another embodiment of the present invention.
  • steps S601 to S611 may correspond to steps S501 to S509 of FIG. 5 described above, and thus may be described with reference to FIG. 5.
  • the service server may compare the number of user-selected images received from the user device and a magnitude relationship between a preset setting value.
  • step S609 If the number of user-selected images is less than the set value, it may be determined that the diagnostic image generated in step S609 does not accurately reflect the user's preference.
  • the user preference information can reflect the user's taste in more detail and accuracy.
  • the user preference information may be extracted by calculating vector values included in fashion items of the user-selected image. Accordingly, as the number of user-selected images increases, user preference information may be extracted through a calculation process in which more vector values are substituted. The more the calculation process is performed, the more sophisticated the user preference information, and the user preference information can reflect the user's taste in more detail and accurately.
  • the service server may extract user preference information based on only one user-selected image.
  • fashion items matched with the preference diagnosis item may be a blue skirt and a red handbag.
  • the service server can extract vector values corresponding to “blue”, “skirt”, “red”, “cross bag”, and “quilt pattern” by combining vector values included in the image of the blue skirt and red handbag.
  • the server can infer that the user prefers fashion items having the characteristics of “blue”, “skirt”, “red”, “cross back”, and “quilt pattern”.
  • weights may be assigned and priority may be determined according to the number of times of overlapping. For example, if the “cross bag” is duplicated 4 times and the “skirt” is duplicated twice, it may be determined that the user's preference for the “cross bag” is higher than that of the “skirt”.
  • the service server may assign more weights to features with high user preference.
  • step S613 it may be assumed that the set value is 5 and the number of user-selected images is 3.
  • the service server may determine that the diagnostic image generated in step S609 does not sufficiently reflect the user's preference information, and perform step S607 to regenerate the diagnostic image.
  • the service server may perform a preference survey again when a user purchases a product different from a preference diagnosis item, which is data for confirming the user's preference.
  • the service server may generate a diagnosis image based on the preference diagnosis item, and update user preference information based on the diagnosis image selected by the user.
  • the service server provides the user with a recommended item determined based on user preference information, and the user may purchase a fashion item different from the recommended item or an item different from the preference diagnosis item provided as a recommendation item.
  • the service server may determine that the previously conducted preference diagnosis is wrong and perform a preference survey again.
  • the process of extracting the preference diagnosis item from the image of the previously purchased item or the possessed item is repeated to regenerate a new preference diagnosis item, or a preference diagnosis item is directly input from the user and diagnosed. You can recreate the image.
  • step S607 the service server informs that the diagnostic image has not sufficiently reflected the user's preference, a message that it is difficult to provide an appropriate recommended item due to insufficient number of user-selected images, or a user selection provided as a precaution.
  • the image is smaller than the set value, a message requesting input of information on the preference diagnosis item together with a message notifying that a new diagnosis image may be provided may be transmitted to the user device.
  • the service server asks the user to take a picture of the preference diagnosis item, requests the input of a keyword that can describe the preference diagnosis item, or requests a voice command through a microphone. You can send a message requesting information.
  • the service server may provide a recommended item to the user device based on the generated user preference information.

Landscapes

  • Business, Economics & Management (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Strategic Management (AREA)
  • Development Economics (AREA)
  • Engineering & Computer Science (AREA)
  • General Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Marketing (AREA)
  • General Physics & Mathematics (AREA)
  • Economics (AREA)
  • Theoretical Computer Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Game Theory and Decision Science (AREA)
  • Data Mining & Analysis (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

L'invention comprend, dans un serveur de service permettant de générer des informations de préférence utilisateur qui sont des informations capables d'estimer la préférence d'un utilisateur pour un article de mode spécifique, une unité de génération d'images de recommandation permettant de générer des images de recommandation pour identifier une préférence de l'utilisateur, d'inclure un élément de recommandation de préférence reflétant les goûts de l'utilisateur, et de fournir les images de recommandation à un dispositif utilisateur à partir d'un carnet de mode comprenant une image de style dans laquelle sont combinés une pluralité d'éléments de mode ; une unité de génération d'informations mises à jour permettant de recevoir une image sélectionnée par l'utilisateur qui est une image sélectionnée par l'utilisateur parmi les images de recommandation, et de générer les informations de préférence de l'utilisateur à partir des éléments de mode inclus dans l'image sélectionnée par l'utilisateur ; et une unité de stockage d'informations de préférence d'utilisateur permettant de stocker les informations de préférence de l'utilisateur.
PCT/KR2020/013648 2019-10-08 2020-10-07 Procédé de recommandation de produit de mode, dispositif et programme informatique Ceased WO2021071238A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
KR10-2019-0124312 2019-10-08
KR1020190124312A KR20210041730A (ko) 2019-10-08 2019-10-08 패션 상품 추천 방법, 장치 및 컴퓨터 프로그램

Publications (1)

Publication Number Publication Date
WO2021071238A1 true WO2021071238A1 (fr) 2021-04-15

Family

ID=75437953

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2020/013648 Ceased WO2021071238A1 (fr) 2019-10-08 2020-10-07 Procédé de recommandation de produit de mode, dispositif et programme informatique

Country Status (2)

Country Link
KR (2) KR20210041730A (fr)
WO (1) WO2021071238A1 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115292534A (zh) * 2022-05-11 2022-11-04 北京五八信息技术有限公司 图像推荐的方法、装置及电子设备

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR102537212B1 (ko) * 2022-02-25 2023-05-26 주식회사 엔엔엠 사용자 정보를 이용한 의류 추천 서비스 제공 장치, 방법 및 프로그램

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20060099377A (ko) * 2005-03-08 2006-09-19 메이크스타일주식회사 소비자의 구매 취향에 따른 상품 검색 서비스 시스템 및방법
KR20120085707A (ko) * 2009-06-03 2012-08-01 라이크닷컴 사용자 장르 및 스타일을 학습하고 사용자 선호도에 따라 상품을 매칭하는 방법 및 시스템
JP2013250743A (ja) * 2012-05-31 2013-12-12 Dainippon Printing Co Ltd コーディネート提案装置、コーディネート提案システム、コーディネート提案方法、プログラム、記録媒体
KR20140075886A (ko) * 2012-12-11 2014-06-20 김주강 코디네이트 배틀 게임을 제공하는 방법 및 서버
KR101762875B1 (ko) * 2016-03-22 2017-07-28 최윤내 큐레이션 서비스를 제공하는 의류 쇼핑 시스템 및 그 방법

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20060099377A (ko) * 2005-03-08 2006-09-19 메이크스타일주식회사 소비자의 구매 취향에 따른 상품 검색 서비스 시스템 및방법
KR20120085707A (ko) * 2009-06-03 2012-08-01 라이크닷컴 사용자 장르 및 스타일을 학습하고 사용자 선호도에 따라 상품을 매칭하는 방법 및 시스템
JP2013250743A (ja) * 2012-05-31 2013-12-12 Dainippon Printing Co Ltd コーディネート提案装置、コーディネート提案システム、コーディネート提案方法、プログラム、記録媒体
KR20140075886A (ko) * 2012-12-11 2014-06-20 김주강 코디네이트 배틀 게임을 제공하는 방법 및 서버
KR101762875B1 (ko) * 2016-03-22 2017-07-28 최윤내 큐레이션 서비스를 제공하는 의류 쇼핑 시스템 및 그 방법

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115292534A (zh) * 2022-05-11 2022-11-04 北京五八信息技术有限公司 图像推荐的方法、装置及电子设备

Also Published As

Publication number Publication date
KR20210041730A (ko) 2021-04-16
KR20220044715A (ko) 2022-04-11

Similar Documents

Publication Publication Date Title
WO2019245316A1 (fr) Système et procédé de génération de recommandations sur la base de descriptions explicables d'aspect amélioré
WO2020085786A1 (fr) Procédé de recommandation de style, dispositif et programme informatique
WO2018225939A1 (fr) Procédé, dispositif et programme informatique pour fournir des publicités basées sur une image
WO2016013914A1 (fr) Procédé, appareil, système et programme d'ordinateur permettant de fournir et d'afficher des informations de produit
WO2020222623A2 (fr) Système et procédé pour construire automatiquement un contenu destiné à des ventes stratégiques
WO2020153796A1 (fr) Dispositif électronique et procédé de fonctionnement associé
WO2019059505A1 (fr) Procédé et appareil de reconnaissance d'objet
WO2021071240A1 (fr) Procédé, appareil et programme informatique pour recommander un produit de mode
WO2022025340A1 (fr) Système de construction de placard virtuel et de création d'une combinaison coordonnée, et procédé associé
WO2020251238A1 (fr) Procédé d'obtention d'informations utilisateur d'intérêt sur la base de données d'image d'entrée et procédé de personnalisation de conception d'objet
WO2017073848A1 (fr) Procédé pour promouvoir ou vendre des produits sur la base d'une activité de service de réseau social de l'utilisateur
WO2021153964A1 (fr) Procédé, appareil et système de recommandation de produit de mode
WO2020251174A1 (fr) Procédé permettant de faire la publicité d'un article de mode personnalisé pour l'utilisateur et serveur exécutant celle-ci
WO2020184855A1 (fr) Dispositif électronique destiné à fournir un procédé de réponse, et son procédé de fonctionnement
WO2021215758A1 (fr) Procédé de publicité pour article recommandé, appareil et programme informatique
WO2021071238A1 (fr) Procédé de recommandation de produit de mode, dispositif et programme informatique
WO2023027224A1 (fr) Procédé et appareil pour fournir des informations de groupage d'éléments
WO2020171567A1 (fr) Procédé permettant de reconnaître un objet et dispositif électronique le prenant en charge
WO2020138941A2 (fr) Procédé de fourniture de service de recommandation d'article de mode à un utilisateur au moyen d'un geste de glissement
WO2020141802A2 (fr) Procédé pour fournir un service de recommandation d'article de mode à un utilisateur en utilisant une date
WO2021040256A1 (fr) Dispositif électronique et son procédé de recommandation de vêtements
WO2022005158A1 (fr) Dispositif électronique et procédé de commande de dispositif électronique
WO2023008617A1 (fr) Système de préparation automatique de bannière publicitaire pour centre commercial en ligne et procédé associé
WO2014119873A1 (fr) Procédé de génération et d'identification d'un code d'identification en forme d'image préférée et procédé pour fournir des informations l'utilisant
WO2024025220A1 (fr) Système pour fournir une plateforme de contenu publicitaire en ligne

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20874704

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20874704

Country of ref document: EP

Kind code of ref document: A1