User portrait generation method, device, computer equipment and storage medium

Abstract

The embodiment of the present application discloses a method, device, computer equipment and storage medium for generating a user portrait. A specific implementation of the method includes: obtaining a user image; inputting the user image into a neural network model for recognition to obtain the user's facial attribute information and / or body attribute information; based on at least one of the preset mapping relationships between facial attribute information and occupational information, facial attribute information and personality information, body attribute information and occupational information, and body attribute information and personality information, obtain the predicted value of the user's occupational information and / or personality information, and generate a user portrait. This implementation can realize the prediction of hidden attribute information such as the user's occupation and personality, thereby realizing a complete user portrait, and combined with data statistics and analysis, more effectively maintain and manage user data.

Description

Technical Field

[0001] The present application relates to the field of Internet technology, and more specifically, to a method, device, computer equipment and storage medium for generating a user portrait. Background Art

[0002] As an effective tool for outlining target users and linking user demands with design directions, user portraits have been widely used in various fields. In the actual operation process, the user's attributes, behaviors and expected data are often transformed and linked in the most simple and life-like words. By building a dedicated user portrait for the user, the user's attributes, preferences, behaviors and other information are displayed, so as to provide different services to the user in a targeted manner.

[0003] At present, the process of user profiling is mainly based on manual review, which builds the user profile based on user information. This manual review method is prone to too many subjective factors in the review process, making it impossible to objectively build the user profile, resulting in inaccurate and incomplete user profiles, which is not conducive to the management of user data in the later stage. Summary of the invention

[0004] The purpose of this application is to provide a user portrait generation method, apparatus, computer device and storage medium to solve at least one of the problems existing in the prior art.

[0005] In order to achieve the above objectives, this application adopts the following technical solutions:

[0006] The first aspect of the present application provides a method for generating a user portrait, comprising:

[0007] Get user image;

[0008] Inputting the user image into a neural network model for recognition to obtain facial attribute information and / or body attribute information of the user;

[0009] Based on at least one of the preset mapping relationships among facial attribute information and occupational information, facial attribute information and personality information, body attribute information and occupational information, and body attribute information and personality information, a predicted value of the user's occupational information and / or personality information is obtained according to the user's facial attribute information and / or body attribute information, and a user portrait is generated.

[0010] The user portrait generation method provided in the first aspect of the present application can predict the user's occupation, personality and other hidden attribute information based on the facial attribute information and / or body attribute information extracted through the neural network model and the preset mapping relationship, thereby obtaining a complete user portrait and user data statistics, improving the accuracy of user portrait construction, and thus improving the accuracy of the services provided to the user, and more effectively maintaining and managing subsequent user data. For example, in the field of smart finance, the user's financial management risk and purchasing power can be predicted and analyzed based on the facial attribute information, body attribute information and the predicted hidden attribute information such as occupation and personality.

[0011] In one possible implementation, the neural network model is a convolutional neural network model.

[0012] In one possible implementation, the convolutional neural network model is a lightweight network model.

[0013] This implementation method uses a lightweight network model to further optimize the calculation of the neural network model while ensuring high calculation accuracy, reducing network parameters, reducing calculation complexity, improving the calculation efficiency of the neural network model, reducing calculation delays, and increasing running speed.

[0014] In a possible implementation, the acquiring the user image includes: acquiring a user face image and / or a user body image;

[0015] The step of inputting the user image into a neural network model for recognition to obtain facial attribute information and / or body attribute information of the user comprises: inputting the user facial image into a neural network model for recognition to obtain facial attribute information of the user and / or inputting the user body image into a neural network model for recognition to obtain body attribute information of the user.

[0016] In a possible implementation, the facial attribute information includes at least one of gender information, expression information, age information, whether the person wears glasses, and whether the person has a beard, and the body attribute information includes hairstyle information and / or clothing information.

[0017] A second aspect of the present application provides a user portrait generation device, comprising:

[0018] An acquisition module, used to acquire user images;

[0019] A neural network model is used to recognize an input user image and obtain the user's facial attribute information and / or body attribute information;

[0020] A prediction module is used to obtain a predicted value of the user's occupational information and / or personality information based on the user's facial attribute information and / or body attribute information, and generate a user portrait based on at least one of a preset mapping relationship between facial attribute information and occupational information, a mapping relationship between facial attribute information and personality information, a mapping relationship between body attribute information and occupational information, and a mapping relationship between body attribute information and personality information.

[0021] In one possible implementation, the neural network model is a convolutional neural network model.

[0022] In one possible implementation, the convolutional neural network model is a lightweight network model.

[0023] In a possible implementation, the acquisition module is used to acquire a user face image and / or a user body image;

[0024] The neural network model is used to recognize the user's facial image to obtain the user's facial attribute information and / or to recognize the user's body image to obtain the user's body attribute information.

[0025] In a possible implementation, the device further includes: a first image collector for collecting a user's facial image and / or at least one second image collector for collecting a user's body image.

[0026] In a possible implementation, the facial attribute information includes at least one of gender information, expression information, age information, whether the person wears glasses, and whether the person has a beard, and the body attribute information includes hairstyle information and / or clothing information.

[0027] The third aspect of the present application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method provided in the first aspect of the present application is implemented.

[0028] The fourth aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method provided in the first aspect of the present application.

[0029] The beneficial effects of this application are as follows:

[0030] In response to the technical problems existing in the prior art, the present application provides a user portrait generation method, device, computer equipment and storage medium. Based on the facial attribute information and / or body attribute information extracted through a neural network model, and a preset mapping relationship, it can realize predictive analysis of the user's occupation, personality, and even hidden attributes such as financial risks and purchasing power, so as to obtain a complete user portrait and user data statistics, improve the accuracy of user portrait construction, and then improve the accuracy of the services provided to users. Moreover, the subsequent user data can be maintained and managed more effectively. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The specific implementation methods of the present application are further described in detail below in conjunction with the accompanying drawings.

[0032] Figure 1 A flowchart of a method for generating a user portrait in one embodiment of the present application is shown.

[0033] Figure 2 An example diagram showing a method for generating a user portrait in one embodiment of the present application.

[0034] Figure 3 A flowchart of neural network model identification in one embodiment of the present application is shown.

[0035] Figure 4 A schematic diagram showing the structure of a user portrait generating device in one embodiment of the present application is shown.

[0036] Figure 5 A schematic diagram showing the structure of a computer system that implements the apparatus provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0037] In order to more clearly illustrate the present application, the present application is further described below in conjunction with the embodiments and drawings. Similar components in the drawings are represented by the same reference numerals. It should be understood by those skilled in the art that the content specifically described below is illustrative rather than restrictive, and should not be used to limit the scope of protection of the present application.

[0038] In the description of the present application, it should be noted that the terms "upper", "lower", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present application. Unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be a connection between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances.

[0039] It should also be noted that, in the description of the present application, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or equipment. In the absence of further restrictions, the elements defined by the statement "comprise one..." do not exclude the presence of other identical elements in the process, method, article or equipment including the elements.

[0040] In view of the technical problems existing in the prior art, an embodiment of the present application provides a method for generating a user portrait. The method can be applied to various servers, computer devices and terminals, and the execution subject of the method can be a processor in the server, computer device or terminal. In a specific example, the method is applicable to the field of smart financial banking system to build an independent user portrait for the user. Similarly, the method can also be applied to other fields where personalized services can be customized, such as the field of automobile sales store systems.

[0041] like Figure 1 As shown, the method for generating a user portrait provided in this embodiment includes the following steps:

[0042] S101, obtaining a user image;

[0043] In one embodiment, obtaining a user image includes obtaining a user face image and / or a user body image, wherein a face image may be an image including the outline of a person's facial features; a body image may be an image including the limbs, torso, clothing, and hairstyle of a person. When obtaining a user image, the execution subject (server or computer device) executing the user portrait generation method may obtain the original information of the target user to be profiled in advance in order to construct a multi-dimensional system user portrait, and the original information includes the personal information of the target user, such as the user's name, native place, permanent residence, place of birth, work information, and other aspects of the user's information.

[0044] In a specific example, the execution entity (server or computer device) that executes the user portrait generation method includes a camera that captures user images (such as photos or video streams), and the user images are captured locally in real time by the camera; in another specific example, user images stored in another server can also be remotely retrieved; in another specific example, user images can also be received from a user's terminal (such as a mobile phone, electronic computer device, etc.) Uploaded.

[0045] Please refer to Figure 2 , Figure 2 An example diagram of a method for generating a user portrait is shown. First, the execution subject (server or computer device) of the method for generating a user portrait includes a camera for collecting user images. After acquiring the user image through the camera, face detection and / or body detection are performed respectively according to the acquired user face image and user body image to determine whether the image contains a face and / or a body, and obtain a user portrait image to be identified and / or a user body image to be identified. Subsequently, it is determined whether the face of the user portrait image to be identified is blocked and whether the face is facing forward to ensure the clarity of the face image. If the face image is unclear, it is prompted that it is unsuccessful, and the user image is acquired again or the next frame is performed.

[0046] Subsequently, a corresponding algorithm is used to perform face recognition on the face image to be identified, that is, face feature extraction; at the same time, the execution subject (server or computer device) of the user portrait generation method retrieves the face features of registered users in the database and compares them with the face features of the face image to be identified to determine whether the user has been registered.

[0047] If after comparison, the user does not exist in the database, the executing entity (server or computer device) executing the user portrait generation method displays a prompt such as "Not registered, please register first" to remind the user to enter original information, such as the user's name, native place, permanent residence, place of birth, work information, etc. After the entry is completed, the executing entity (server or computer device) executing the user portrait generation method assigns a new user identity ID to the user based on the original information provided by the user, and binds the user's original information, extracted facial features, and acquired user facial images and user body images to the user identity ID as explicit attributes in the user's user portrait.

[0048] If after comparison, there is a user in the database whose facial features match the facial image to be identified, the executing entity (server or computer device) that executes the user portrait generation method displays the user's identity ID for the user to confirm, and at the same time binds the facial features of the newly added facial image in this collection to the user's identity ID.

[0049] S102, inputting the user image into the neural network model for recognition, and obtaining the user's facial attribute information and / or body attribute information;

[0050] In an embodiment where acquiring a user image includes acquiring a user facial image and / or a user body image, inputting the user image into a neural network model for recognition to obtain the user's facial attribute information and / or body attribute information includes: inputting the user facial image into a neural network model for recognition to obtain the user's facial attribute information and / or inputting the user's body image into a neural network model for recognition to obtain the user's body attribute information.

[0051] In a specific embodiment, the face attribute information includes at least one of gender information, expression information, face shape information, age information, whether to wear glasses, and whether to have a beard; in a specific example, the face shape information may include: round face, square face, triangle face, melon-shaped face, heart-shaped face, etc. In addition, the body attribute information includes hairstyle information and / or clothing information. In another specific example, the hairstyle information may include straight, wavy, curly, braided, bun or short hair, etc. Clothing information may include formal wear, business, casual, sports or others.

[0052] In a specific embodiment, the neural network model is a convolutional neural network model, and the user's face image and / or user's body image are input into the convolutional neural network model, thereby extracting the feature vector of the user's face image and / or user's body image, the feature vector is a vector used to represent the features of the face image and / or body image, the feature vector of the face image is different from the feature vector of the body image, a preset number of face images and / or body images are input into the convolutional neural network model, and the convolutional neural network model outputs the feature vector of each face image or body image, so that the feature vector of each face image or body image can be obtained. The convolutional neural network model can be ResNet50 (ResidualNetwork50, residual network model 50), FPN (Feature Pyramid Networks, feature pyramid network model), GoogleNet (Google Networks, a network model), ShuffleNet (Shuffle Networks, convolutional compression network model), or a network model composed of ResNet50 and FPN.

[0053] Considering that an overly complex network model will slow down the calculation speed, and in order to optimize the calculation efficiency of the traditional convolutional neural network, the accuracy of the model calculation is guaranteed while maintaining the speed of the model calculation, ensuring timely feedback and low latency. In a specific embodiment, the convolutional neural network model is a lightweight network model. This embodiment can be applied to banking scenarios or 4S scenarios where there is a large flow of traffic, a small amount of single data processing, and personalized customizable services that require fast feedback.

[0054] In a further embodiment, the lightweight network model may be a ShuffleNet_v2 model, which can effectively reduce network parameters by adopting different convolution calculation methods, thereby simplifying the convolution calculation of the traditional convolutional neural network model and improving its computational efficiency, thereby achieving the best model accuracy using limited computing resources.

[0055] Compared with other lightweight network models, the ShuffleNet_v2 model has the advantages of lower computational complexity and higher accuracy in addition to its fast running speed. It is particularly suitable for running in mobile terminals with limited computing resources. For example, it can be installed in the mobile terminals of operators or marketers, which makes it convenient for operators to carry it with them for real-time observation and analysis.

[0056] This embodiment uses a lightweight network model to further optimize the calculation of the neural network model while ensuring high calculation accuracy, reducing network parameters, reducing calculation complexity, improving the calculation efficiency of the neural network model, reducing calculation delay, and increasing the running speed. At the same time, the ShuffleNet_v2 model can be used to set the neural network model in the mobile terminal, which is convenient for operators to carry with them, view user attributes and analysis in real time, and improve the efficiency of providing personalized services to different users.

[0057] See also Figure 3 , Figure 3 A flowchart of neural network model identification according to an embodiment.

[0058] First, we use datasets with different facial attribute information (including gender information, expression information, face shape information, age information, whether wearing glasses, whether having a beard, etc.) and datasets with different human body attribute information (including hairstyle and clothing type) for model training.

[0059] In such Figure 3 In the illustrated embodiment, feature extraction is performed by inputting the user's face image and / or the user's body image into a trained feature extraction model, and then a pooling operation is performed, wherein the pooling operation refers to the aggregation statistics of the 512-dimensional vector data extracted by the feature extraction model. In a specific example, the pooling operation is an average pooling operation, that is, the maximum value is extracted from each row (or column) of the output matrix to represent the row (or column). Under the premise of retaining the overall data characteristics, the dimension of the features extracted by the neural network model is effectively reduced, and better features with stronger information are extracted at the same time, which reduces the difficulty of calculation and speeds up the calculation.

[0060] The features after the pooling layer are input into the classifier (SoftMax) and the regularization model respectively. The features after the pooling layer can be used by the classifier to obtain some of the user's facial attribute information and body attribute information, such as gender information, expression information, face shape information, whether wearing glasses, whether having a beard, hairstyle information and clothing information. In addition, the features after the pooling layer can be used by the regularization model, that is, local response normalization, to obtain some of the user's facial attribute information, such as age information. The regularization model is to introduce additional information into the original model. By adopting the regularization model, the risk of overfitting can be effectively reduced and the generalization performance of the neural network model can be improved.

[0061] In a further embodiment, the regularized model adopts L1 norm regularization, that is, by adding the L1 norm (L1 norm refers to the sum of the absolute values ​​of each element in the vector data) to the original model, the output result satisfies sparsity, further reducing the risk of overfitting, improving the generalization performance of the neural network model, and facilitating feature extraction.

[0062] S103. Based on at least one of the preset mapping relationships among facial attribute information and occupational information, facial attribute information and personality information, body attribute information and occupational information, and body attribute information and personality information, a predicted value of the user's occupational information and / or personality information is obtained according to the user's facial attribute information and / or body attribute information, and a user portrait is generated.

[0063] In a specific example, personality information may include rational, doubtful, introverted, extroverted, and mixed; and occupational information may include business talent, high-salary talent, and salaried talent.

[0064] In a specific embodiment, the mapping relationship between the facial attribute information, the body attribute information and the user's occupation information and personality information may be pre-built and saved, for example, including the following steps:

[0065] Construct an input value set X, i.e., facial attribute information and body attribute information, including, for example: gender (male, female), face shape (round face, square face, triangular face, oval face, heart-shaped face), glasses (wearing glasses, not wearing glasses), beard (with beard, without beard), hairstyle (straight, wavy, curly, braid, bun, short hair), clothing (formal suit, business suit, casual wear, sportswear, other).

[0066] Construct the output value set Y, for example: personality (rational, doubtful, introverted, extroverted, mixed), occupation (business talent, high-paid talent, salaried talent).

[0067] Set the function f (denoted as f:X→Y) so that the input value set X and the output value set Y satisfy a many-to-many mapping relationship and satisfy the following conditions: for each element x in the input value set X, there is an element y in the output value set Y, satisfying that x and y are related by f.

[0068] For example, the mapping function f1-f6 is as follows:

[0069] f1: X(beard) → Y(personality) = {with beard → rational, introvert, extrovert; without beard → rational, introvert, extrovert, mixed};

[0070] f2: X(glasses) → Y(personality) = {wearing glasses → rational, doubtful, introverted, extroverted; not wearing glasses → rational, introverted, extroverted, mixed};

[0071] f3: X (hair style) → Y (personality) = {straight → rational, doubtful, introverted; wavy → rational, introverted, extroverted; curly → rational, doubtful, introverted, extroverted, mixed; braided → extroverted, mixed, bun, short → rational, doubtful, introverted, extroverted, mixed};

[0072] f4: X (hair style) → Y (occupation) = {straight → high-paid talents, salaried talents; wavy → business talents, high-paid talents; curly → high-paid talents, salaried talents; braid → high-paid talents, salaried talents, short hair → business talents, high-paid talents, salaried talents};

[0073] f5: X(clothing)→Y(personality)={formal wear→rational, doubtful, introverted; business wear→rational, introverted, extroverted, mixed; casual wear→doubtful, extroverted, mixed; sportswear→rational, extroverted, mixed; others→rational, doubtful, introverted, extroverted, mixed};

[0074] f6: X(clothing)→Y(occupation)={Formal wear→high-paid talents, salaried talents; Business wear→high-paid talents, salaried talents; Casual wear→high-paid talents, salaried talents; Sports wear→salaried talents, others→salaried talents}.

[0075] According to any of the above mapping relationships between the facial attribute information, body attribute information and the user's occupational information and personality information, the predicted value of the user's occupational information and / or personality information is obtained according to the user's facial attribute information and / or body attribute information. For example, if the user's body attribute information is: the hairstyle information is straight, according to the mapping function f3, the predicted value of the user's personality information is: rational type, doubtful type, introverted type.

[0076] In a specific embodiment, multiple mapping relationships between facial attribute information, body attribute information and the user's occupational information and personality information are used to predict the user's occupational information and / or personality information, and the final prediction result is obtained based on the intersection of the multiple mapping relationships.

[0077] In a specific example, if the user's facial attribute information is: beard, glasses; the body attribute information is: wavy hairstyle, business suit. According to the mapping function f1-f6, the predicted values ​​of the user's occupation information and personality information are: rational, introverted; the occupation information is: high-paid talent.

[0078] In a specific example, the predicted values ​​of the user's occupational information and / or personality information and the user's facial attribute information and / or body attribute information are uniformly output for attribute statistics, such as statistics on the gender distribution, age distribution, face shape distribution, appearance distribution, user distribution of different hairstyles, user distribution of different clothing, predicted occupation distribution, predicted personality type distribution, financial risk type distribution, purchasing power distribution, number of registered users, etc. of daily visiting users, and user information analysis is performed to facilitate operators and marketing personnel to intuitively understand user information, and effectively provide different services or products for users with different financial risks or different purchasing powers. At the same time, by merging the predicted values ​​of the user's occupational information and / or personality information and the user's facial attribute information and / or body attribute information with the original information of the user previously obtained, extracting facial features and assigning a new user identity ID, a complete user portrait for the user is formed, thereby improving the accuracy of user portrait construction, improving the accuracy of services or products provided to the user, and facilitating the maintenance and management of subsequent user data.

[0079] The user portrait generation method provided in this embodiment is based on facial attribute information such as gender / age of the face and / or body attribute information such as hairstyle / clothing type extracted through a neural network model, and a preset mapping relationship. It can realize predictive analysis of hidden attributes such as the user's occupation, personality, and even financial risk and purchasing power, thereby obtaining a complete user portrait and user data statistics, improving the accuracy of user portrait construction, and then improving the accuracy of the services provided to users. In addition, subsequent user data can be maintained and managed more effectively.

[0080] Please refer to Figure 4 Another embodiment of the present application provides a user portrait generation device 200. The embodiment of the user portrait generation device 200 is similar to Figure 1 Corresponding to the embodiment of the user portrait generation method shown, the user portrait generation device 200 can be specifically applied to various servers, computer devices and terminals. The user portrait generation device 200 includes an acquisition module 201, which is used to acquire a user image; a neural network model 202, which is used to recognize the input user image and obtain the user's facial attribute information and / or body attribute information; a prediction module 203, which is used to obtain the predicted value of the user's occupation information and / or personality information based on the user's facial attribute information and / or body attribute information based on at least one of the preset mapping relationship between facial attribute information and occupation information, the mapping relationship between facial attribute information and personality information, the mapping relationship between body attribute information and occupation information, and the mapping relationship between body attribute information and personality information, and generate a user portrait.

[0081] In a specific embodiment, the user portrait generating device 200 includes a front end and a cloud end, wherein the front end is used to collect user portraits (such as user facial portraits and / or user body portraits) and obtain the user's original information, extract facial features, and assign a new user identity ID, and transmit the above information to the cloud end for statistics and analysis; the cloud end is used to identify the user facial portrait and / or user body portrait, obtain the user's facial attribute information and / or body attribute information, and predict the identified facial attribute information and / or body attribute information, and perform statistics on the user's facial attributes and / or body attributes and predicted values, for example, statistics are obtained on the gender distribution, age distribution, face shape distribution, appearance distribution, distribution of users with different hairstyles, distribution of users with different clothing, predicted occupational distribution, predicted personality type distribution, financial risk type distribution, purchasing power distribution, number of registered users, etc. of daily visiting users, and perform user information analysis, and form a complete user portrait by merging it with the user's original information obtained by the front end, extracting facial features, and assigning a new user identity ID, and then maintaining and managing subsequent user data.

[0082] In another specific example, the front end is only used to collect user portraits (such as user facial portraits and / or user body portraits); and the cloud is also used to obtain the user's original information, extract facial features, and assign a new user identity ID, that is, in this embodiment, the cloud is used to build a complete user portrait of the user.

[0083] This embodiment operates through the integration of the front-end and the cloud, and is suitable for the user portrait system in the field of smart finance, and can realize flexible deployment of the user portrait construction process.

[0084] In a specific embodiment, the neural network model 202 is a convolutional neural network model. In a further embodiment, the convolutional neural network model is a lightweight network model.

[0085] In a specific embodiment, the acquisition module 201 is used to acquire a user's facial image and / or a user's body image; the neural network model 202 is used to recognize the user's facial image to obtain the user's facial attribute information and / or recognize the user's body image to obtain the user's body attribute information.

[0086] In a specific embodiment, the user portrait generation device 200 also includes: a first image collector for collecting user facial images and / or at least one second image collector for collecting user body images. The user facial images and user body images can be either pictures or video streams. In the example applied to the field of smart financial banking systems, the first image collector can be a camera set on an ATM machine, and the second image collector can be at least one surveillance camera set in a bank lobby. In addition, the user facial image and the user body image can also be collected simultaneously by only at least one surveillance camera set in the bank lobby, thereby reducing hardware facilities. When the clarity meets the requirements, the user facial image and the user body image only need to be obtained from one camera at the same time, which can effectively improve the processing efficiency.

[0087] In a specific embodiment, the face attribute information includes at least one of gender information, expression information, face shape information, age information, whether to wear glasses, and whether to have a beard; in a specific example, the face shape information may include: round face, square face, triangular face, melon-shaped face, heart-shaped face, etc. In addition, the body attribute information includes hairstyle information and / or clothing information. In another specific example, the hairstyle information may include straight, wavy, curly, braided, bun or short hair, etc. Clothing information may include formal wear, business, casual, sports or others.

[0088] It should be noted that the principle and workflow of the user portrait generation device 200 provided in this embodiment are similar to those of the above-mentioned user portrait generation method. The relevant parts can be referred to the above description and will not be repeated here.

[0089] like Figure 5 As shown, another embodiment of the present application provides a computer device suitable for implementing the user portrait generation method provided in the above embodiment, including a central processing module (CPU), a memory, and a computer program stored in the memory and running on the central processing module (CPU), which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) or the program loaded from the storage part into the random access memory (RAM). In the random access memory (RAM), various programs and data required for the operation of the computer system are also stored. The CPU, ROM, and RAM are connected to each other via a bus. The input / output (I / O) interface is also connected to the bus.

[0090] The following components are connected to the I / O interface: an input part including a keyboard, a mouse, etc.; an output part including an organic light emitting diode display (OLED) or an inorganic light emitting diode display (LED), etc. and a speaker, etc.; a storage part including a hard disk, etc.; and a communication part including a network interface card such as a LAN card, a modem, etc. The communication part performs communication processing via a network such as the Internet. The drive is also connected to the I / O interface as needed. Removable media, such as a disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., are installed on the drive as needed so that the computer program read therefrom is installed into the storage part as needed.

[0091] In particular, according to this embodiment, the process described in the flowchart above can be implemented as a computer software program. For example, this embodiment includes a computer program product, which includes a computer program tangibly contained on a computer-readable medium, and the computer program includes program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication part, and / or installed from a removable medium.

[0092] The flow chart and schematic diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product of the present embodiment. In this regard, each square box in the flow chart or schematic diagram can represent a part of a module, program segment or code, and a part of the above-mentioned module, program segment or code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the schematic diagram and / or flow chart, and the combination of the square boxes in the schematic diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0093] The modules described in this embodiment may be implemented by software or hardware. The modules described may also be arranged in a processor, for example, may be described as: a processor including an acquisition module, a recognition module and a prediction module. The names of these modules do not constitute limitations on the modules themselves in certain circumstances. For example, the recognition module may also be described as a "feature extraction module".

[0094] As another aspect, this embodiment further provides a non-volatile computer storage medium, which may be the non-volatile computer storage medium included in the above-mentioned device in the above-mentioned embodiment, or may be a non-volatile computer storage medium that exists independently and is not installed in the terminal. The above-mentioned non-volatile computer storage medium stores one or more programs, and when the above-mentioned one or more programs are executed by a device, the above-mentioned device implements the method for generating a user portrait as provided in the above-mentioned embodiment.

[0095] Obviously, the above embodiments of the present application are merely examples for clearly illustrating the present application, and are not limitations on the implementation methods of the present application. For ordinary technicians in this field, other different forms of changes or modifications can be made based on the above description. It is impossible to list all the implementation methods here. All obvious changes or modifications derived from the technical solution of the present application are still within the scope of protection of the present application.

Claims

1. A method for generating a user portrait, characterized in that: include: Get user image; Inputting the user image into a neural network model for recognition to obtain facial attribute information and / or body attribute information of the user; Based on at least one of a preset mapping relationship between facial attribute information and occupation information, a mapping relationship between facial attribute information and personality information, a mapping relationship between body attribute information and occupation information, and a mapping relationship between body attribute information and personality information, a predicted value of the user's occupation information and / or personality information is obtained according to the user's facial attribute information and / or body attribute information, and a user portrait is generated; The user portrait generation method also includes: merging the predicted value of the user's occupational information and / or personality information, as well as the user's original information, extracting facial features, and assigning a new user identity ID to form a complete user portrait of the user.

2. The method according to claim 1, characterized in that The neural network model is a convolutional neural network model.

3. The method according to claim 2, characterized in that The convolutional neural network model is a lightweight network model.

4. The method according to claim 1, characterized in that The acquiring of the user image comprises: acquiring a user face image and / or a user body image; The step of inputting the user image into a neural network model for recognition to obtain facial attribute information and / or body attribute information of the user comprises: inputting the user facial image into a neural network model for recognition to obtain facial attribute information of the user and / or inputting the user body image into a neural network model for recognition to obtain body attribute information of the user.

5. The method according to claim 1, characterized in that: The facial attribute information includes at least one of gender information, expression information, age information, whether the person wears glasses, and whether the person has a beard, and the body attribute information includes hairstyle information and / or clothing information.

6. A user portrait generation device, characterized in that: include: An acquisition module, used to acquire user images; A neural network model is used to recognize an input user image and obtain the user's facial attribute information and / or body attribute information; A prediction module, for obtaining a predicted value of the user's occupation information and / or personality information according to the user's facial attribute information and / or body attribute information, and generating a user portrait based on at least one of a preset mapping relationship between facial attribute information and occupation information, a mapping relationship between facial attribute information and personality information, a mapping relationship between body attribute information and occupation information, and a mapping relationship between body attribute information and personality information; The prediction module is also used to merge the predicted values ​​of the user's occupation information and / or personality information, the user's original information, extract facial features, and assign a new user identity ID to form a complete user portrait of the user.

7. The device according to claim 6, characterized in that The neural network model is a convolutional neural network model.

8. The device according to claim 7, characterized in that The convolutional neural network model is a lightweight network model.

9. The device according to claim 6, characterized in that The acquisition module is used to acquire a user face image and / or a user body image; The neural network model is used to recognize the user's facial image to obtain the user's facial attribute information and / or to recognize the user's body image to obtain the user's body attribute information.

10. The device according to claim 9, characterized in that The device also includes: a first image collector for collecting a user's face image and / or at least one second image collector for collecting a user's body image.

11. The device according to claim 6, characterized in that The facial attribute information includes at least one of gender information, expression information, age information, whether the person wears glasses, and whether the person has a beard, and the body attribute information includes hairstyle information and / or clothing information.

12. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 5 is implemented.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images