JP2009128995A - Merchandise affinity calculation method, program and device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To support a user to purchase accessories through a network without imposing any large labor on a customer side by: analyzing the images of target accessories; and automatically performing ideal combination. <P>SOLUTION: A combination decision system 5 calculates a distance (difference) between a feature value received from a fashion selling system 3 and the feature value of each merchandise registered in a merchandise feature value database 7, and calculates the affinity of merchandise designated by a user and the other merchandise based on the distance. By referring to the affinity of merchandise B to the merchandise A transmitted to a user terminal 1, the user is able to apply for the purchase of merchandise B whose affinity to the merchandise A is satisfactory, and which seems to be used after combined with the merchandise A with satisfactory affinity after purchase to the fashion selling system 3. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a merchandise compatibility degree calculation method, program, and apparatus, and more particularly to a technique in which a user searches for a merchandise having a high compatibility degree of a combination of a plurality of merchandise.

  In Patent Literature 1, for each product, product information data D1 indicating the product name and price, image data D2, and product coordination data D3 indicating association with other products that can be coordinated are prepared, and each customer is provided. , Past purchase history data D4, customer image data D5 composed of the customer's own face photo, and customer information data D6 indicating the address and name are prepared. When the customer selects the first desired product by the selection unit, the selection unit automatically selects other products that can be coordinated with reference to the data D3 and D4, and selects the second desired product from among them. . The image composition unit presents the customer with a try-on image obtained by compositing the face photograph and the product. A flyer on which the second desired product is posted and a purchase application form are created by the flyer creation section and the application form creation section, and transmitted online by the information transmission section.

  In Patent Document 2, a consumer connects to a mediator's home page, selects a product registered in a server as a registered virtual try-on model, and makes a virtual try-on. In virtual try-on, advice can be obtained with diagnostic software created based on the data of the color coordinator and stylist, and if virtual try-on is performed, the manufacturer name, purchased product name, price, etc. are displayed in a list, and the consumer provides ordering information. Then, the broker sends the consumer data necessary for the contract to each seller over the Internet, and the seller sends the goods to the consumer.

In Patent Document 3, it is possible to detect that clothes are tagged and the clothes are removed from the storage. Using this information, the user profile is updated. In this way, the system can learn from user behavior and generate more accurate recommendations over time. The user profile can be stored on a smart card, so that the user can use the smart card in the shop to purchase new clothing for his set. The user profile can also be used for targeted advertising.
JP 2004-295777 A JP 2001-216429 A JP-T-2004-507822

  When purchasing clothing, bags, shoes, accessories, sunglasses, and other clothing items, the compatibility with the products purchased at the same time is an important point in determining the purchase of clothing items. When purchasing at an actual store, it can be confirmed with the advice and try-on of the store clerk, but when purchasing through the Internet, it is necessary for the user to imagine and judge, so there is a possibility that many users will not be able to complete the purchase. It is high, and it seems that the market e-commerce rate is low compared to books.

  The present invention supports the purchase of clothing via the user's network without imposing a heavy burden on the seller site by analyzing the image of the target clothing and automatically performing an ideal combination. For the purpose.

  The present invention is a merchandise compatibility degree calculation method for a computer to calculate a compatibility degree, which is a numerical value indicating the compatibility of a combination of merchandise items related to a decorative product, and is two desired merchandise composing a desired merchandise combination. Based on the step of designating the first product and the second product, the step of calculating the feature quantity of the first product based on the content of the image of the first product, the content of the image of the second product, Calculating a feature amount of the second product, and calculating a degree of compatibility between the first product and the second product based on a distance between the feature amount of the first product and the feature amount of the second product. And outputting the calculated degree of compatibility between the first product and the second product.

  According to the present invention, the compatibility of an arbitrary combination of products is indicated by an objective numerical value called compatibility, so an end user who wants to purchase a product refers to the compatibility even without expert knowledge about coordination. Can grasp the ideal combination of products.

  Preferably, the method includes a step of receiving at least one of an image of the first product and an image of the second product from the user terminal.

  Preferably, the method includes a step of selecting, from the user terminal, at least one of an image of the first product and an image of the second product from the images of the sales product registered in advance in a predetermined sales site.

  Preferably, the degree of compatibility is output to the user terminal.

  Preferably, the feature amount of the first product is calculated based on the content of the image of the first product and the incidental information of the image of the first product, and the feature amount of the second product is the second product's feature amount. It is calculated based on the content of the image and the incidental information of the image of the second product.

  In this way, information that cannot be grasped from the content of the image itself, such as trends and manufacturers, can be reflected in the feature amount.

  Preferably, one or a plurality of ideal product sets and a feature value of a product composing the set are registered in the database, and a product having a feature value closest to the feature value of the first product and ideal One of the products constituting the typical product set, the first similar product, and the product having the feature quantity closest to the feature quantity of the second product, and constituting the ideal product set Extracting a second similar product that is a product of the first product from the database, a first reference product that is the other product that forms an ideal product set together with the first similar product, and a second similar product And extracting a second reference product, which is the other product that constitutes the ideal product set, from the database, and the compatibility of the first product and the second product is determined based on the first product Features and the first reference product Distance symptoms amount, and is calculated based on the length of the second feature quantity and the feature quantity of the second reference items of goods.

  A program for causing a computer to execute the above-described merchandise compatibility degree calculation method is also included in the present invention.

  The present invention is a merchandise compatibility degree calculation device for calculating a degree of compatibility, which is a numerical value indicating the compatibility of a combination of merchandise items related to a decorative product, and is a first two desired merchandise items constituting a desired merchandise combination. Based on the content of the image of the first product, the means for calculating the feature quantity of the first product based on the content of the image of the first product, Means for calculating the feature quantity of the first product, means for calculating the degree of compatibility between the first product and the second product based on the distance between the feature quantity of the first product and the feature quantity of the second product, Means for outputting the calculated degree of compatibility between the first product and the second product.

  According to the present invention, the compatibility of an arbitrary combination of products is indicated by an objective numerical value called compatibility, so an end user who wants to purchase a product refers to the compatibility even without expert knowledge about coordination. Can grasp the ideal combination of products.

  FIG. 1 is a schematic configuration diagram of a clothing item purchase support system according to a preferred embodiment of the present invention. This system includes input operation means (keyboard, mouse, touch panel, etc.), display means (liquid crystal display, etc.), communication means (network card, etc.), information processing means (CPU, microcomputer, etc.), storage means (RAM, ROM, hard disk, CD-RW and other removable recording media), user terminal 1, fashion sales system composed of a personal computer equipped with all or part of information reading / writing means (CD-RW drive, etc.) 3, a feature amount calculation system 4, a combination determination system 5, a product image database 6, a product feature amount database 7, a user information database 8, a seller terminal 9, and a product image registration system 10. The user terminal 1 and the fashion sales system 3 are connected via a network 2 such as the Internet.

  The fashion sales system 3 is connected to the feature amount calculation system 4, the combination determination system 5, the user information database 8, and a LAN or other communication means.

  The combination determination system 5 is connected to the product feature amount database 7 via a LAN or other communication line.

  Seller terminal 9 and product image registration system 10, product image registration system 10 and feature quantity calculation system 4 and product image database 6, feature quantity calculation system 4 and product feature quantity database 7, combination determination system 5 and product feature quantity database 7 Are connected by a LAN or other communication line.

  The user terminal 1 can communicate with other personal computers, mobile phones, and other information processing terminals, and has a display function.

  The network 2 is configured by the Internet or the like, but may be another network.

  The fashion sales system 3 accepts a function for electronically selling apparel (by e-commerce), that is, a request for selling a desired apparel from the user terminal 1, and associates information related to the accepted product with user information. This is a system for registering in the information database 8.

  The feature amount calculation system 4 includes image information sent by the user via the user terminal 1 or product information registered in the product image database 6 and additional information (tag information) for each product image input from the seller terminal 9. This is a system for calculating the feature quantity from and registering it in the product feature quantity database 7.

  The feature amount is a numerical value obtained by analyzing an image, and is a numerical value corresponding to the feature as a clothing item or a combination thereof. For example, the overall color (R, G, B expression), the color for each part (for example, the color for each area of the upper, middle, lower, left and right sides), the overall shape (jacket, coat, sweater, polo shirt, etc.) , Neck shape (V-neck, U-neck, round neck, etc.), chest shape (starched buzzam, pleated buzzam, etc.), pattern (houndstooth check, leopard print, etc.), contents of tag information attached to the image ( Product price, brand, trend) etc. are included in whole or part. The method of digitizing the features is arbitrary. For colors, for example, R = 1, G = 2, B = 3, and the like. Alternatively, the neck shape is V neck = 1, U neck = 2, round neck = 3, and the like. In short, numerical values and features correspond to each other on a one-to-one basis, and similar values may be given a close value, and non-similar features may be given a non-close value.

  The combination determination system 5 is a system that calculates the compatibility of the combination of products based on the distance between the feature amount of the product selected by the user terminal 1 and the feature amount registered in the product feature amount database 7.

  The product image database 6 is a database in which images, prices, product numbers, etc. of products handled by the fashion sales system 3 are registered.

  The product feature quantity database 7 is a database in which feature quantities for each unspecified number of products calculated by the feature quantity calculation system 4 for the fashion sales system 3 are registered. The range of product feature amounts to be registered is arbitrary, and may be limited to products sold by a specific manufacturer, or all products handled by the fashion sales system 3 may be included.

  The user information database 8 is a product specified by the user via the user terminal 1 as a target for combination determination (only one product of the combination may be specified or both may be specified), whether or not there is a purchase, This is a database in which attribute information such as history, past determination results of the combination determination system 5, a set of popular products, user sex, age, and body type are registered.

  The seller terminal 9 is a terminal for registering tag information that the business operator of the fashion sales system 3 wants to reflect in the feature amount of the product image.

  The product image registration system 10 is a system for adding tag information describing additional information to a product image and registering it in the product image database 6. The tag information includes product price, brand, trendy color, trendy style, and the like.

  Hereinafter, the flow of processing performed in the present system will be described with reference to FIG.

  First, the user accesses the fashion sales site established by the fashion sales system 3 from the user terminal 1 constituted by a PC, a mobile phone, etc. (S1), and designates a product for which the combination is determined (S2). This product may be a product sold by the fashion sales system 3 or a product already owned by the user. When designating a product owned by the user, the image is uploaded to the fashion sales system 3. In addition, what is necessary is just to use what was image | photographed with the digital camera or the mobile phone with a camera for the image of goods.

  The fashion sales system 3 inputs an image of the product designated by the user to the feature amount calculation system 4 (S3). However, this is not essential, and the product image registered in the fashion sales system 3 or an image downloaded from another site is selected from the user terminal 1, and the selected image is input to the feature amount calculation system 4. May be.

  The feature amount calculation system 4 calculates a feature amount for the input or selected image and returns it to the fashion sales system 3 (S4).

  The fashion sales system 3 transmits the feature amount received from the feature amount calculation system 4 to the combination determination system 5 (S5).

  The combination determination system 5 calculates the distance (difference) between the feature quantity received from the fashion sales system 3 and the feature quantity of each product registered in the product feature quantity database 7 (S6), and based on this distance. The degree of compatibility between the product designated by the user and the other product is calculated (S7).

  In order to extract a product that is compatible with the product arbitrarily designated by the user from the sales products registered in the fashion sales system 3, the following is performed.

  That is, the feature quantity of a basic product group (standard items such as white shirts, bag blazers, etc.) is calculated in advance, and one or more products that are compatible with each basic product are selected. A specialist such as a stylist selects in advance as an ideal combination, and the ideal combination (for example, the product A1 and the product C1, or the product B1 and the product D1) is used as a combination (learning data) of the compatibility 100. And registered in the product feature quantity database 7. At that time, the feature quantity of the product for each product pair (for example, product A1, product B1, product C1, product D1) constituting an ideal combination is also included in the learning data and stored. The learning data is preferably created for products of various genres. For example, even for combinations using the same product, it is preferable to prepare several types such as an ideal combination for work and an ideal combination for everyday wear.

  For example, when the combination determination system 5 searches for a product B that is compatible with an arbitrary product A specified by the user, first, the product having the smallest difference in feature value compared to the feature values of the products A and B (Similar products) is extracted from the product feature database 7. For the convenience of explanation, among the feature quantities registered in the product feature quantity database 7, the similar product having the smallest feature value difference from the feature quantity of the product A is the product A1, and the feature quantity of the product B It is assumed that the similar product having the smallest feature value difference value is the product B1.

  Next, the combination determination system 5 extracts, from the product feature amount database 7, the other product (reference product) having the similar product A1 and the similar product B1 as one of a pair of ideal combinations (compatibility 100). For convenience of explanation, in the set registered in the product feature quantity database 7, the other product that makes a pair of the product A1 and the ideal set is the product C1, and makes a pair of the product B1 and the ideal set. Assume that the other product is the product D1.

  Finally, the compatibility between the product A and the product B is calculated based on the similarity between the feature values of the product A and the reference product C1 and the similarity of the feature values between the product B and the reference product D1. The calculation of the compatibility degree is based on the compatibility Z of the products A and B. For example, the compatibility Z = 100−α × (| feature amount of the product A−feature amount of the product C1 | + | feature amount of the product B−feature amount of the product D1 |). α is a ratio for converting the feature quantity into a compatibility degree that takes a range of 0 to 100. If the feature quantities of the products A and C1 are approximated and the feature quantities of the products B and D1 are approximated, the degree of compatibility Z becomes closer to 100 accordingly. On the other hand, if the feature quantities of the products A and C1 are not approximated or the feature quantities of the products B and D1 are not approximated, the compatibility Z is small. The product B is an arbitrary product registered in the fashion sales system 3, and the compatibility Z between the product A and the product B is calculated for each product B.

  The combination determination system 5 returns each of the compatibility calculated for each registered product and the product designated by the user to the combination determination system 5 together with the identification information of each corresponding product (S8). The identification information of each product and the compatibility corresponding to each product are referred to as compatibility information.

  The fashion sales system 3 registers all or part of the compatibility information received from the combination determination system 5 in the user information database 8 (S9). Since the low compatibility degree has a low utility value, for example, only compatibility degree information having the top ten compatibility degrees may be registered.

  The fashion sales system 3 extracts product identification information from the compatibility information registered in the user information database 8 (S10). Next, the product image database 6 is searched using the extracted identification information as a key. And the product image matched with the said identification information is extracted from the product image database 6, and the extracted product image is transmitted to the user terminal 1 via the network 2 (S11) with the corresponding compatibility degree (S12). .

  The user refers to the degree of compatibility of the product B with the product A sent to the user terminal 1, and purchases the product B that is compatible with the product A and can be used in combination with the product A after purchase. Can be applied to the fashion sales system 3.

  Various usages of the user of this system can be assumed. For example, the user can search for a product that is compatible with the clothes on hand. In this case, both the products A and B may be specified from the user terminal 1.

  Or, when users make bulk purchases, look at the degree of compatibility between the products they want to purchase, and select the ideal purchase variation for a specialized coordination that will increase the number of combinations that increase the degree of compatibility. It can be formulated without knowledge.

  Alternatively, the user should prioritize the use of clothing items (such as wedding receptions, parties, interviews, commuting, everyday wear) and product attributes such as fashion, color, shape, etc. If it is specified not only to be related but also to tag information, it is sufficient if it is related to the feature quantity), the degree of compatibility that weights the condition or the corresponding feature quantity is calculated, and the user needs You can propose a combination that suits your needs.

  Or the compatibility degree with respect to the combination of the goods which the user coordinated can be calculated, and objective determination can also be given to a user's coordination. For example, a set (α, β) of clothing images coordinated by the user himself (which may be an image on the sales site or an image taken by the user) is prepared, and one of the products α constituting the set , Search for compatible products. Then, the similarity between the product δ extracted as a result of the search and the product β selected by the user is determined based on the distance between the feature amounts of both products, and the similarity is pointed to the user terminal 1. Present. If the point is high, it can be seen that the coordinates determined by the user are objectively compatible.

  Alternatively, a product similar to the product designated by the user can be searched. For example, the user terminal 1 provides the fashion sales system 3 with an image of a product (which may be an image in a sales site or an image taken by the user) as a search key, and the fashion sales system 3 uses the image as a feature amount. It sends to the calculation system 4 to calculate the feature amount. Next, the product feature quantity database 7 sequentially extracts the products having the feature quantity and the feature quantity having the smallest distance. It can be said that the degree of similarity with the specified product is higher as the distance of the extracted feature amount is smaller. This can be used when the user searches for a product that is similar in design but cheaper.

Block diagram of an accessory purchase support system according to an embodiment of the present invention Conceptual diagram of compatibility calculation

Explanation of symbols

  1: user terminal, 3: fashion sales system, 4: feature quantity calculation system, 5: combination determination system, 6: product image database, 7: product feature quantity database, 8: user information database, 9: seller terminal, 10 : Product image registration system

Claims (8)

A product compatibility degree calculation method for a computer to calculate a compatibility degree, which is a numerical value indicating a compatibility of a combination of products related to an ornament,
Designating a first product and a second product, which are two desired products constituting a combination of desired products;
Calculating the feature quantity of the first product based on the content of the image of the first product;
Calculating a feature amount of the second product based on the content of the image of the second product;
Calculating the degree of compatibility between the first product and the second product based on the distance between the feature value of the first product and the feature value of the second product;
Outputting the calculated degree of compatibility between the first product and the second product;
Product compatibility calculation method including   The merchandise compatibility degree calculation method according to claim 1, further comprising: receiving at least one of an image of the first product and an image of the second product from a user terminal.   The method according to claim 1, further comprising: selecting, from a user terminal, at least one of an image of the first product and an image of the second product from an image of the sales product registered in advance in a predetermined sales site. The product compatibility degree calculation method described.   The commodity compatibility degree calculation method according to claim 2 or 3, wherein the compatibility degree is output to the user terminal. The feature amount of the first product is calculated based on the content of the image of the first product and the incidental information of the image of the first product,
The product compatibility degree according to any one of claims 1 to 4, wherein the feature amount of the second product is calculated based on content of an image of the second product and incidental information of the image of the second product. Calculation method. Registering one or a plurality of ideal product sets and feature quantities of the products constituting the set in a database;
Features of a first similar product, which is a product having a feature quantity that most closely approximates the feature quantity of the first product, and is one of the products constituting the ideal product set, and features of the second product Extracting from the database a second similar product that is a product having a feature quantity closest to the quantity and is one of the products constituting the ideal product set;
The first reference product, which is the other product that constitutes the ideal product set together with the first similar product, and the other product that constitutes the ideal product set together with the second similar product Extracting a second reference product that is from the database;
Including
The degree of compatibility between the first product and the second product is the distance between the feature value of the first product and the feature value of the first reference product, and the feature value of the second product and the first value. The merchandise compatibility degree calculating method according to claim 1, wherein the merchandise compatibility degree is calculated based on a distance between feature quantities of two reference merchandise.   The program for making a computer perform the merchandise compatibility degree calculation method in any one of Claims 1-6. A product compatibility degree calculation device for calculating a compatibility degree, which is a numerical value indicating a compatibility of a combination of products related to an ornament,
Means for designating a first product and a second product which are two desired products constituting a combination of desired products;
Means for calculating a feature quantity of the first product based on the content of the image of the first product;
Means for calculating a feature quantity of the second product based on the content of the image of the second product;
Means for calculating a degree of compatibility between the first product and the second product based on a distance between the feature value of the first product and the feature value of the second product;
Means for outputting the calculated degree of compatibility between the first product and the second product;
Product compatibility calculation device including

Images