A virtual fitting room implementation method based on user model and collaborative filtering algorithm

By establishing a virtual model of users and products, using virtual fitting rooms and collaborative filtering algorithms, the problem that users cannot try products in online shopping is solved, the shopping experience and recommendations are improved, and the return rate and waste of transportation resources are reduced.

CN114862486BActive Publication Date: 2024-11-01QINGDAO INST OF COMPUTING TECH XIDIAN UNIV +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202110154316.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-02-04
Publication Date
2024-11-01
Estimated Expiration
2041-02-04

AI Technical Summary

Technical Problem

Because online shopping platforms cannot provide the effect of users trying products on their own, they have difficulty in choosing, high return rates, waste of transportation resources, and lack of scientific product recommendations.

Method used

By establishing a user's three-dimensional body model and product parameter model, using a virtual fitting room and collaborative filtering algorithm, users can experience products, and recommend products based on user feedback and interest similarity.

Benefits of technology

It improves user shopping experience, reduces selection difficulties and return rates, and achieves more objective and data-based product recommendations.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114862486B_ABST
    Figure CN114862486B_ABST
Patent Text Reader

Abstract

The present invention belongs to the field of multimedia information processing. It improves the user's purchasing experience and performs similarity recommendation through a network virtual fitting room and a collaborative filtering algorithm. Specifically, it is a virtual fitting room implementation method based on a user model and a collaborative filtering algorithm. The method collects the user's three-dimensional body data and the attribute data of the commodity and stores them in a user database and a commodity database respectively, and uses 3dmax to model the user and the commodity. When the user selects a specific commodity, the commodity model with the corresponding attribute is found from the database and combined with the user model for imaging. After that, the user can use "like" or "dislike" to give feedback on the commodity. After obtaining the user's feedback information, the system will make recommendations based on the user's preferences. The present invention conveniently completes the use and feedback of commodities through a virtual fitting room, thereby improving the user's purchasing experience.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of multimedia information processing, and improves the user's purchasing experience and recommendation through a network virtual fitting room and a collaborative filtering algorithm. Background Art

[0002] At present, the display and promotion of goods on online shopping platforms are completed by hiring third-party professional organizations. Since users cannot see the effects of using the goods like buying goods in physical stores, and due to the limitations of space and time, users can only use the goods in the mall or store where they are located, and cannot choose other goods for trial, which limits their purchases. At the same time, there are no system recommendations, so there is often difficulty in choosing and even the possibility of giving up buying. In terms of transportation costs, since users lack direct observation of the effects of using the goods, they apply for returns if the goods do not meet the users' expectations, resulting in a waste of logistics and transportation resources. Due to the lack of understanding of user preferences and the similarity of goods, store owners can only evaluate and recommend based on their own intuition, lacking certain data statistics and scientificity, which often makes users more confused and gives up buying. Summary of the invention

[0003] The present invention aims to solve at least one of the above problems to a certain extent. To enable users to have a better shopping experience and to understand the effects of using the goods in more detail, the present invention is based on a reliable character model established by the user and a 1:1 model established based on the parameters of the goods, so that users can conveniently use the goods and provide feedback through a virtual fitting room, so that the system can better understand the user's preferences and complete the similarity calculation of the goods and make reasonable recommendations, narrowing the user's selection range and improving the user's purchasing experience.

[0004] The present invention is achieved in that the specific steps include:

[0005] Step 1: Collect accurate data of the user's three-dimensional body including height, gender, weight, arm length, leg length, chest circumference, waist circumference, hip circumference, and non-beautified photos to provide skin color parameters, etc., to provide accurate digital basis for the user's model;

[0006] Step 2: Use 3dmax to build a 1:1 model based on the collected 3D body data of the user;

[0007] Step 3: Collect the product attribute parameters including shoulder width, chest circumference, waist circumference, leg circumference, length and style, etc., to provide accurate numerical basis for the product model;

[0008] Step 4: Build a 1:1 model based on the collected commodity attribute information;

[0009] Step 5: Combine the user model and the product model selected by the user to form an image;

[0010] Step 6: The system will find a set of users with similar interests based on the user's feedback information; given user a and user b, let f(a) represent the set of items that user a has given positive feedback on, and f(b) represent the set of items that user b has given positive feedback on, then according to the Jaccard formula

[0011]

[0012] or

[0013]

[0014] w ab Indicates the similarity of interests between user a and user b. Since there are popular items, both user a and user b may buy them, which may affect w ab The accuracy of The impact of popular items in the common interest lists of user a and user b on similarity is penalized.

[0015] Furthermore, find items in this collection that the user likes and that the target user has not heard of and recommend them to the user; it is necessary to build an inverted list of items and users, and then build an n*n user similarity matrix W. If user A and user B are both interested in item a, then add 1 to W[A][B] and W[B][A], and so on. After scanning all the items, the final W matrix is ​​obtained. The W matrix is ​​the numerator of the cosine similarity. Dividing W by the denominator can get the final user interest similarity.

[0016] Furthermore, after obtaining the similarity between users, the interest of user u in item i is calculated according to the following formula:

[0017]

[0018] Where S(u, K) contains K users with similar interests to user u, N(i) is the set of users who have taken actions on item i, and w uv represents the similarity between user u and user v, r vi Represents user v’s interest in item i, here the default is r vi =1.

[0019] The specific beneficial effects of the present invention are:

[0020] (1) The present invention simulates the real effect based on the virtual model established by the user and the product, so the user does not need to go to the physical store to experience it but can experience and use the product more conveniently through the virtual fitting room;

[0021] (2) The present invention is based on a reliable character model established by the user and a 1:1 model established based on the parameters of the product. In the traditional shopping method, users can only judge whether they need to buy based on pictures, which may cause the user to purchase inappropriate products and waste human and logistics resources. Therefore, compared with the traditional method, this virtual fitting room is more reliable;

[0022] (3) The present invention completes recommendations based on a collaborative filtering algorithm, which is more objective and data-based than traditional methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a flow chart of a virtual fitting room based on user modeling and collaborative filtering algorithm of the present invention. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0025] The application principle of the present invention is described in detail below in conjunction with the accompanying drawings.

[0026] like Figure 1 As shown, the present invention comprises the following steps:

[0027] Step 1: Collect accurate data of the user's three-dimensional body including height, gender, weight, arm length, leg length, chest circumference, waist circumference, hip circumference, and non-beautified photos to provide skin color parameters, etc., to provide accurate digital basis for the user's model;

[0028] Step 2: Use 3dmax to build a 1:1 model based on the collected 3D body data of the user;

[0029] Step 3: Collect the product attribute parameters including shoulder width, chest circumference, waist circumference, leg circumference, length and style, etc., to provide accurate numerical basis for the product model;

[0030] Step 4: Build a 1:1 model based on the collected product attribute information;

[0031] Step 5: Combine the user model and the product model selected by the user to form an image;

[0032] Step 6: The system will find a set of users with similar interests based on the user's feedback information; given user a and user b, let f(a) represent the set of items that user a has given positive feedback on, and f(b) represent the set of items that user b has given positive feedback on, then according to the Jaccard formula

[0033]

[0034] or

[0035]

[0036] w ab Indicates the similarity of interests between user a and user b. Since there are popular items, both user a and user b may buy them, which may affect w ab The accuracy of The impact of popular items in the common interest lists of user a and user b on similarity is penalized.

[0037] Furthermore, find items in this collection that the user likes and that the target user has not heard of and recommend them to the user; it is necessary to build an inverted list of items and users, and then build an n*n user similarity matrix W. If user A and user B are both interested in item a, then add 1 to W[A][B] and W[B][A], and so on. After scanning all the items, the final W matrix is ​​obtained. The W matrix is ​​the numerator of the cosine similarity. Dividing W by the denominator can get the final user interest similarity.

[0038] Furthermore, after obtaining the similarity between users, the interest of user u in item i is calculated according to the following formula:

[0039]

[0040] Where S(u, K) contains K users with similar interests to user u, N(i) is the set of users who have taken actions on item i, and w uv represents the similarity between user u and user v, r vi Represents user v’s interest in item i, here the default is r vi =1.

[0041] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for implementing a virtual fitting room based on a user model and a collaborative filtering algorithm, characterized in that: The following steps are involved: Step 1: The user inputs the three-dimensional data of the shape and stores it in the user database; Step 2: Collect the three-dimensional data of the product and store it in the product database; Step 3: When the user selects the corresponding product, the three-dimensional model of the corresponding product is read from the database; Step 4: Combine the user model and the product model selected by the user to form an image; Step 5: Users provide feedback using "like" or "dislike" based on their preferences; Step 6: The system calculates similarity and makes recommendations based on user feedback; The system calculates similarity based on user feedback information and recommends the following specific steps: Step 1: The system will find a set of users with similar interests based on the user's feedback information; given user a and user b, let f(a) represent the set of items that user a has given positive feedback on, and f(b) represent the set of items that user b has given positive feedback on, then according to the Jaccard formula or w ab Indicates the similarity of interests between user a and user b. Since there are popular items, both user a and user b may buy them, which may affect w ab The accuracy of Penalizes the impact of popular items in the common interest list of user a and user b on similarity; Step 2: Find items in this collection that the user likes and has not heard of, and recommend them to the user. You need to build an item-user inverted list, and then build an n*n user similarity matrix W. If user A and user B are both interested in item a, add 1 to W[A][B] and W[B][A], and so on. After scanning all items, you will get the final W matrix, which is the numerator of cosine similarity. Dividing W by the denominator will give you the final user interest similarity. Step 3: After obtaining the similarity between users, calculate the interest of user u in item i according to the following formula: Where S(u, K) contains K users with similar interests to user u, N(i) is the set of users who have taken actions on item i, and w uv represents the similarity between user u and user v, r vi Represents user v’s interest in item i, here the default is r vi =1; Step 4: After calculating the interest index of user u for item i, make recommendations based on the order of the index.

2. The method for implementing a virtual fitting room based on a user model and a collaborative filtering algorithm according to claim 1, characterized in that: Establishing a user 3D model specifically includes: Get the three-dimensional shape parameters input by the user; The user's three-dimensional body parameters include: height, gender, weight, arm length, leg length, chest circumference, waist circumference, hip circumference, and non-beautified photos are used to provide skin color parameters; A user 3D model is established based on the 3D body parameters provided by the user.

3. The method for implementing a virtual fitting room based on a user model and a collaborative filtering algorithm according to claim 1, characterized in that: Building a 3D model of a product includes: Get the attribute parameters of the product; The attribute parameters of the product include shoulder width, chest circumference, waist circumference, leg circumference, length and style of the product; Establish a three-dimensional model of the product based on the product's attribute parameters.

4. The method for implementing a virtual fitting room based on a user model and a collaborative filtering algorithm according to claim 1, characterized in that: Obtaining a 3D body model of the user and a 3D model of the product; The user's model is matched one by one with models of different product models and combined into an image.

5. The method for implementing a virtual fitting room based on a user model and a collaborative filtering algorithm according to claim 1, characterized in that: The user's feedback after using the product, and statistics on the feedback results.