CN114201631A

CN114201631A - Photo publishing method and device, computer equipment and storage medium

Info

Publication number: CN114201631A
Application number: CN202010973014.7A
Authority: CN
Inventors: 吴歆婉; 夏胜飞; 宋睿
Original assignee: Tencent Cyber Shenzhen Co Ltd
Current assignee: Tencent Cyber Shenzhen Co Ltd
Priority date: 2020-09-16
Filing date: 2020-09-16
Publication date: 2022-03-18

Abstract

The application relates to a photo publishing method, a photo publishing device, computer equipment and a storage medium. The method comprises the following steps: responding to a photo uploading operation triggered in the social application, displaying the candidate photos, responding to a selection operation of the candidate photos, and acquiring a target photo to be uploaded; entering a photo confirmation page, and displaying the target photo in a state of responding to editing operation; displaying a quality evaluation result of the target photo corresponding to the target photo; the quality evaluation result represents a quality score obtained by carrying out artificial intelligence scoring on the target photo; and in response to a publishing operation triggered by the photo confirmation page, publishing the target photos sorted according to the quality scores to a personal page of the social application when the number of the target photos is at least two. By adopting the method, the photo publishing efficiency can be improved.

Description

Photo publishing method and device, computer equipment and storage medium

Technical Field

The present application relates to the field of computer technologies, and in particular, to a method and an apparatus for publishing a photo, a computer device, and a storage medium.

Background

When a user surfs the internet, the user often publishes a picture shot by the user on the internet so as to attract other users to pay attention. The current software generally provides a beauty function or a photo pendant function and the like so as to meet the requirements of users. The user can beautify the photo on the software, or set an emoticon on the photo and then release the emoticon. However, the current photo distribution method has the problem of low distribution efficiency.

Disclosure of Invention

In view of the above, it is desirable to provide a method, an apparatus, a computer device, and a storage medium for distributing photos, which can improve the efficiency of distributing photos.

A method of publishing a photo, the method comprising:

responding to a photo uploading operation triggered in the social application, displaying the candidate photos, responding to a selection operation of the candidate photos, and acquiring a target photo to be uploaded;

entering a photo confirmation page, and displaying the target photo in a state of responding to editing operation;

displaying a quality evaluation result of the target photo corresponding to the target photo; the quality evaluation result represents a quality score obtained by carrying out artificial intelligence scoring on the target photo;

and in response to a publishing operation triggered by the photo confirmation page, publishing the target photos sorted according to the quality scores to a personal page of the social application when the number of the target photos is at least two.

An artificial intelligence based photo scoring method, the method comprising:

acquiring target facial features of a target photo, and acquiring target photo parameters of the target photo;

processing the target facial features and the target photo parameters respectively by adopting at least two universal attraction models to obtain the similarity of each universal attraction model to the target photo; each universal attraction model corresponds to a different basic score;

screening out a target universal attraction model matched with the target photo according to the similarity;

and generating a quality score of the target photo according to the basic score corresponding to the target universal attraction model.

A photo publication apparatus, the apparatus comprising:

the uploading module is used for responding to photo uploading operation triggered in the social application, displaying the candidate photos and responding to selection operation of the candidate photos to acquire target photos to be uploaded;

the display module is used for entering a photo confirmation page and displaying the target photo in a state of responding to editing operation;

the display module is used for displaying the quality evaluation result of the target photo corresponding to the target photo; the quality evaluation result represents a quality score obtained by carrying out artificial intelligence scoring on the target photo;

and the publishing module is used for responding to a publishing operation triggered by the photo confirmation page, and publishing the target photos sequenced according to the quality scores to the personal page of the social application when the number of the target photos is at least two.

In one embodiment, the display module is used for displaying a photo processing switch item on the photo confirmation page;

responding to the triggering operation of the photo processing switch item, and adjusting the state of the photo processing switch item to be one of an opening state and a closing state;

the publishing module is used for responding to a publishing operation triggered by the photo confirmation page when the photo processing switch item is in an on state, and publishing the target photos sequenced according to the quality scores to the personal page of the social application when the number of the target photos is at least two.

In one embodiment, the target photos displayed in the photo confirmation page are arranged in the selected order; the publishing module is used for responding to a publishing operation triggered on the photo confirmation page when the photo processing switch item is in a closed state, and publishing the target photos to the personal page of the social application according to the sequence arranged in the photo confirmation page when the number of the target photos is at least two.

In one embodiment, the publishing module is configured to display a target photo with the highest quality score on a cover of the private page in response to a publishing operation triggered on the photo confirmation page when the photo processing switch item is in an on state.

In one embodiment, the editing operation comprises a replacement operation. The uploading module is used for responding to replacement operation triggered by the target photo displayed in the photo confirmation page, displaying the candidate photo again, responding to selection operation of the displayed photo again, and acquiring the target photo to be replaced; when the photo confirmation page is returned, replacing the target photo acted by the replacement operation with the target photo to be replaced; and the display module is used for displaying the quality evaluation result of the replaced target photo corresponding to the replaced target photo.

In one embodiment, the display module is used for displaying prompt information representing a scoring process corresponding to the target photo when the target photo is subjected to artificial intelligence scoring;

and after the artificial intelligence scoring of the target photo is finished, displaying a quality evaluation result in a character expression form converted from the quality scoring of the target photo corresponding to the target photo.

An artificial intelligence based photograph scoring apparatus, the apparatus comprising:

the characteristic acquisition module is used for acquiring target facial characteristics of a target photo and acquiring target photo parameters of the target photo;

the similarity obtaining module is used for processing the target facial features and the target photo parameters respectively by adopting at least two universal attraction models to obtain the similarity of each universal attraction model to the target photo; each universal attraction model corresponds to a different basic score;

the screening module is used for screening out a target general attraction model matched with the target photo according to the similarity;

and the quality score generation module is used for generating the quality score of the target photo according to the basic score corresponding to the target general attraction model.

In one embodiment, the feature acquisition module is configured to acquire a target photo and convert the target photo into a gray-scale image; carrying out face recognition based on the gray level image to obtain a face area in the target picture; extracting facial feature points from the facial region; and determining feature points and facial features of the target photo based on the facial feature points.

In one embodiment, the feature point facial features comprise a facial global feature, a facial distance feature and a facial area feature;

the feature acquisition module is used for performing combined calculation on feature point coordinates corresponding to the facial feature points to obtain facial feature vectors, and the facial feature vectors are used for representing facial global features;

determining the sizes of organs in the face and the distances among the organs based on the feature point coordinates to obtain face distance features;

and determining the area surrounded by the facial feature points based on the feature point coordinates, determining the area of each area based on the feature point coordinates, and normalizing the area to obtain facial area features.

In one embodiment, the quality score generation module is used for obtaining the maximum additional score of the target universal attraction model;

determining the additional score of the target photo according to the similarity corresponding to the target universal attraction model and the maximum additional score;

and obtaining the quality score of the target photo according to the basic score and the additional score corresponding to the target universal attraction model.

In one embodiment, the artificial intelligence-based photo scoring device comprises a training module, wherein the training module is used for acquiring at least two sample photos and quality scoring labels corresponding to the classified sample photos, and the categories of the sample photos are used for representing different basic scores;

extracting sample facial features and sample photo parameters of the sample photo;

processing the samples based on the facial features and the photo parameters of the samples in the corresponding categories respectively by adopting at least two untrained universal attraction models to obtain the similarity of the untrained universal attraction models to the target photos in the corresponding categories; the untrained universal attraction model corresponds to a base score;

obtaining a sample quality score of the sample photo according to a basic score corresponding to the untrained universal attraction model and by combining the similarity;

and adjusting parameters in the untrained universal attraction model according to the sample quality score and the quality score label, and continuing training until a trained universal attraction model is obtained.

In one embodiment, the training module is used for acquiring at least two sample photos and interaction data corresponding to each sample photo from a social application;

classifying the sample photos based on the interaction data according to the preset number of untrained universal attraction models to obtain classified sample photos, wherein the preset number is at least two;

and carrying out normalization processing on each interactive data to obtain a quality score mark corresponding to each classified sample photo.

In one embodiment, the training module is configured to, for each sample photo, obtain the like number corresponding to each of the at least two like types;

and carrying out weighting processing on the corresponding praise number based on the weight corresponding to each praise type to obtain interactive data.

In one embodiment, the target facial features include at least one of global facial features, facial distance features, facial area features, facial appearance features, hair style features, facial expression features; the target photo parameters comprise at least one of resolution, brightness, sharpness, contrast, overall tone, size and position relation between a human body and the photo, shooting place and shooting time.

A computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:

A computer-readable storage medium, on which a computer program is stored which, when executed by a processor, carries out the steps of:

According to the photo publishing method, the device, the computer equipment and the storage medium, artificial intelligent scoring is carried out on the target photo uploaded by the user to obtain a quality scoring result, and the quality scoring result is displayed, so that the quality of the target photo uploaded by the user can be objectively reflected; in response to the publishing operation triggered by the photo confirmation page, when the number of the target photos is at least two, the target photos sequenced according to the quality scores are published to the personal page of the social application, namely the target photos sequenced according to the quality scores can be displayed on the personal page of the social application, the position of the photos does not need to be manually adjusted by a user, and the photo publishing efficiency is improved.

Drawings

FIG. 1 is a flowchart illustrating a method for publishing a photo in one embodiment;

FIG. 2 is a schematic diagram of an interface for displaying candidate photos, according to an embodiment;

FIG. 3 is a schematic interface diagram of a photo confirmation page in one embodiment;

FIG. 4 is a schematic interface diagram of a personal page of a social application in one embodiment;

FIG. 5 is a schematic interface diagram of a photo confirmation page in accordance with an alternative embodiment;

FIG. 6 is a schematic diagram of an interface of a personal page of a social application in another embodiment;

FIG. 7 is a schematic flow chart diagram illustrating an artificial intelligence based photo scoring method in accordance with an embodiment;

FIG. 8 is a schematic diagram of facial distance features in one embodiment;

FIG. 9 is a schematic illustration of a facial distance feature in another embodiment;

FIG. 10 is a schematic illustration of facial area features in one embodiment;

FIG. 11 is a schematic flow chart diagram illustrating an artificial intelligence photo scoring method in accordance with another embodiment;

FIG. 12 is a flow diagram illustrating a manner in which a generic attraction model may be trained in one embodiment;

FIG. 13 is a flowchart illustrating a training method of the generic attraction model according to another embodiment;

FIG. 14 is a schematic flow chart diagram illustrating an artificial intelligence based photo scoring method in accordance with yet another embodiment;

FIG. 15 is a block diagram showing the construction of a photograph issuing apparatus in one embodiment;

FIG. 16 is a block diagram of an artificial intelligence based photo scoring apparatus in one embodiment;

FIG. 17 is a diagram illustrating an internal structure of a computer device according to an embodiment.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.

The photo publishing method can be only applied to the terminal device, and can also be applied to an application scene comprising the terminal device and the server. The terminal device communicates with the server through the network. The terminal device may be: the intelligent terminal comprises intelligent terminals such as a smart phone, a tablet computer, a notebook computer, a desktop computer and an intelligent television. The terminal equipment is provided with a client of the social application, and the client of the social application can be used for publishing photos, texts and the like. The server may be an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, or a cloud server providing basic cloud computing services such as cloud service, a cloud database, cloud computing, a cloud function, cloud storage, network service, cloud communication, middleware service, domain name service, security service, CDN, and a big data and artificial intelligence platform. The number of the terminal devices and the number of the servers are not limited. The terminal equipment responds to a photo uploading operation triggered in the social application, displays the candidate photo, responds to a selection operation of the candidate photo, and acquires a target photo to be uploaded; entering a photo confirmation page, and displaying the target photo in a state of responding to the editing operation; displaying a quality evaluation result of the target photograph corresponding to the target photograph; the quality evaluation result represents a quality score obtained by carrying out artificial intelligence scoring on the target photo; and in response to a publishing operation triggered by the photo confirmation page, publishing the target photos sorted according to the quality scores to a personal page of the social application when the number of the target photos is at least two.

In an embodiment, as shown in fig. 1, a photo publishing method is provided, which is described by taking an example that the method is applied to a terminal device, and includes the following steps:

and 102, responding to a photo uploading operation triggered in the social application, displaying the candidate photo, responding to a selection operation of the candidate photo, and acquiring a target photo to be uploaded.

The social application refers to application software for a user to perform social contact. The social application may be, for example, a WeChat, QQ, or the like application. The photo uploading operation refers to an operation for uploading a photo to a server corresponding to the social application. The photo uploading operation can be triggered by the photo uploading control element in the touch terminal device. The candidate photo may be a photo local to the terminal device, or a photo directly obtained from a website by the terminal device, or a photo taken by the terminal device. The target photo to be uploaded is at least one of the candidate photos.

Specifically, the terminal device displays a photo uploading interface of the social application, and the photo uploading interface comprises a photo uploading control. And the terminal equipment responds to the photo uploading operation triggered in the social application and displays the candidate photos. And the terminal equipment responds to the selection operation of the candidate photo to acquire the target photo to be uploaded. For example, the candidate photos include photo a, photo B, photo C, and photo D, and the target photos to be uploaded may be photo a, photo B, and photo C.

And 104, entering a photo confirmation page and displaying the target photo in a state of responding to the editing operation.

The photo confirmation page is a page for confirming a target photo to be uploaded. The state in response to the editing operation means a state in which the operation such as replacement or deletion can be responded. Prompt information indicating the scoring process may be included in the photo confirmation page.

Specifically, the photos are uploaded to a server corresponding to the social application in response to a confirmation operation on the photo uploading page. And the terminal equipment enters a photo confirmation page, and displays the target photo in the response editing state on the photo confirmation page.

And 106, displaying the quality evaluation result of the target photo corresponding to the target photo. The quality evaluation result represents a quality score obtained by artificially and intelligently scoring the target photo.

The quality evaluation result can be embodied directly through the quality score, can also be embodied through the quality evaluation grade, and can also be embodied through the ranking of the photo quality in the social application without being limited to the above. The quality score may be, for example, 90 points, 89.5 points, etc. The quality evaluation grade can be grade A, grade B, etc. The ranking of the photos in the social application may be "you beat 70% of net friends" or "your score beats 1234567 people", etc.

Artificial Intelligence (AI) is a theory, method, technique and application system that uses a digital computer or a machine controlled by a digital computer to simulate, extend and expand human Intelligence, perceive the environment, acquire knowledge and use the knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technique of computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a manner similar to human intelligence. Artificial intelligence is the research of the design principle and the realization method of various intelligent machines, so that the machines have the functions of perception, reasoning and decision making.

The artificial intelligence technology is a comprehensive subject and relates to the field of extensive technology, namely the technology of a hardware level and the technology of a software level. The artificial intelligence infrastructure generally includes technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and the like.

Specifically, the terminal device uploads the target photo to the server, and the server performs artificial intelligence scoring on the photo. Or the terminal equipment carries out artificial intelligence scoring on the photos. The quality score is generated according to a basic score corresponding to the target universal attraction model, and the target universal attraction model is screened out according to the similarity of each universal attraction model to the target photo; the similarity of each attraction model corresponding to the target photo is obtained by processing at least two general attraction models respectively based on the sample facial features and the sample photo parameters, wherein each general attraction model corresponds to different basic scores. The terminal device may display the quality evaluation result of the target photograph in a display area of the target photograph, or may display the quality evaluation result and the like in a lower area of the target photograph, corresponding to the target photograph.

And step 108, in response to the publishing operation triggered by the photo confirmation page, publishing the target photos sorted according to the quality scores to a personal page of the social application when the number of the target photos is at least two.

The ranking according to the quality scores may be ranking in order from top to bottom according to the quality scores, or may be ranking by only displaying the photos with the highest quality scores at the top. The private page is used to display personal information including, but not limited to, published photos, user nicknames, user graduate schools, etc.

Specifically, in response to a publishing operation triggered on the photo confirmation page, when the number of the target photos is at least two, the terminal device publishes the target photos sorted according to the quality scores to a private page of the social application. The target photos ordered by the quality scores can be displayed on the personal page of the social application. For example, a photos score 70 points, B photos score 50 points, and C photos score 90 points, then C photos, a photos, and B photos are displayed in order on the social application's personal page.

According to the photo publishing method, artificial intelligence scoring is carried out on the target photo uploaded by the user to obtain a quality scoring result, and the quality scoring result is displayed, so that the quality of the target photo uploaded by the user can be objectively reflected; in response to the publishing operation triggered by the photo confirmation page, when the number of the target photos is at least two, the target photos sequenced according to the quality scores are published to the personal page of the social application, that is, the target photos sequenced according to the quality scores can be displayed on the personal page of the social application, the position of the photos does not need to be manually adjusted by a user, the photo publishing efficiency is improved, the user can obtain more attention in the social application, and the user experience is improved.

In one embodiment, the photo publishing method further comprises: displaying a photo processing switch item on a photo confirmation page;

and when the photo processing switch item is in an on state, executing a step of responding to a publishing operation triggered on the photo confirmation page, and publishing the target photos sequenced according to the quality scores to a personal page of the social application when the number of the target photos is at least two.

The photo processing switch item can be used for confirming whether at least one of the target photos are ranked according to the quality scores obtained by the artificial intelligence score and whether the target photos with the highest quality scores are displayed on the cover page of the personal page.

Specifically, a photo process switch item is displayed on the photo confirmation page. And responding to the triggering operation of the photo processing, and adjusting the state of the photo processing switch item to be in an opening state. Or, in response to a trigger operation for photo processing, adjusting the state of the photo processing switch item to an off state. The photo confirmation page includes a photo publication control. The publication operation can be triggered through the photo publication control. And when the photo processing switch item is in an on state, executing a step of responding to a publishing operation triggered on the photo confirmation page, and publishing the target photos sequenced according to the quality scores to a personal page of the social application when the number of the target photos is at least two.

According to the photo publishing method, the photo confirmation page comprises the photo processing switch item, when the photo processing switch item is in an open state, the target photos sequenced according to the quality scores are published to the personal page of the social application, so that a user can independently select whether intelligent sequencing is needed, when the photo processing switch item is selected to be opened, the user does not need to manually adjust the positions of the photos, and the photo publishing efficiency is improved.

In one embodiment, the target photos displayed in the photo confirmation page are arranged in the selected order; the photo publishing method further comprises the following steps:

and when the photo processing switch item is in a closed state, in response to a publishing operation triggered on the photo confirmation page, publishing the target photos to the private page of the social application in the order arranged in the photo confirmation page when the target photos are at least two.

Specifically, the target photos displayed in the photo confirmation page are arranged in the order selected by the user. And when the photo processing switch item is in a closed state, responding to the issuing operation triggered on the photo confirmation page, and when the number of the target photos is at least two, the terminal equipment issues the target photos to the personal page of the social application according to the sequence arranged in the photo confirmation page. The target photos are displayed in the order they are arranged in the confirmation page in the personals page of the social application. For example, if the order selected by the user is photograph C, photograph B, and photograph A, then the order in the photograph confirmation page is photograph C, photograph B, and photograph A.

According to the photo publishing method, when the photo processing switch items are in the closed state, the target photos are published in the personal pages applied by the social contact according to the sequence arranged in the photo confirmation page under the condition that the target photos are at least two, and then a user can independently select whether intelligent sequencing is needed or not, so that the user experience is improved.

In one embodiment, the photo publishing method further comprises: and when the photo processing switch item is in an on state, displaying the target photo with the highest quality score on the cover page of the personal page in response to the publishing operation triggered on the photo confirmation page.

Wherein the cover sheet refers to the content displayed first in the private page. When the second user accesses the first user's personal page, the first user's personal cover is displayed first.

Specifically, when the photo process switch item is in the on state, the terminal device displays, on the cover of the private page, the highest-quality-score target photo among the target photos in response to the posting operation triggered on the photo confirmation page.

In this embodiment, the photo publishing method further includes: and when the photo processing switch item is in an opening state, responding to the release operation triggered on the photo confirmation page, comparing the first quality score of the target photo with the highest quality score with the second quality score of the released photo in the personal cover, and displaying the photo with the highest quality score in the first quality score and the second quality score on the cover of the personal page.

According to the photo publishing method, when the photo processing switch item is in the open state, the target photo with the highest quality score is displayed on the cover of the personal page in response to the publishing operation triggered by the photo confirmation page, so that the attraction of the user in social application can be improved, and more attention can be obtained.

In one embodiment, the editing operation includes a replacing operation, and the photo publishing method further includes:

responding to a replacement operation triggered by a target photo displayed in the photo confirmation page, displaying the candidate photo again, responding to a selection operation of the photo displayed again, and acquiring the target photo to be replaced;

when the picture confirmation page is returned, replacing the target picture acted by the replacement operation with the target picture to be replaced;

and displaying the quality evaluation result of the replaced target photo corresponding to the replaced target photo.

Wherein, the replacing operation refers to replacing one photo with another photo. Displaying the candidate photos again in response to the replacement operation triggered by the target photo displayed in the photo confirmation page; and in response to the selection operation of the photo displayed again, the target photo to be replaced is acquired. And after the picture confirmation page is returned, the terminal equipment replaces the target picture acted by the replacing operation with the target picture to be replaced. And displaying the quality evaluation result of the replaced target photo on the photo confirmation page corresponding to the replaced target photo. For example, when the quality evaluation result of the displayed target photograph includes a negative evaluation result, for example, the negative evaluation result is "photo change bar", and the target photograph corresponding to the negative evaluation result is target photograph C. The target photo responds to a replacement operation triggered by the target photo C displayed in the photo confirmation page, displays the candidate photo A, the candidate photo B and the candidate photo C again, and responds to a selection operation of the displayed photo again to obtain the target photo B to be replaced; and when the picture confirmation page is returned, replacing the target picture C with the target picture B, and displaying the quality evaluation result of the target picture B.

According to the photo publishing method, the target photo acted by the replacing operation is replaced by the target photo to be replaced in response to the replacing photo triggered by the target photo, the quality evaluation result of the replaced target photo is still displayed for the replaced target photo, the photo with poor quality can be replaced when the quality evaluation result is poor, and the quality of the published photo is improved.

In one embodiment, the displaying the quality evaluation result of the target photo corresponding to the target photo comprises: when the target photo is subjected to artificial intelligence scoring, displaying prompt information representing a scoring process corresponding to the target photo; and after the artificial intelligence scoring of the target photo is completed, displaying a quality evaluation result in a character expression form converted from the quality scoring of the target photo corresponding to the target photo.

The prompt information in the scoring process may be represented in the form of a progress bar, a progress percentage, or a text prompt information, for example, but is not limited thereto. The quality assessment results in the form of textual expression may be a spoken description, such as "you are more than XX% of the users," which may be calculated as a percentage of people who are more than the attraction model.

Specifically, when the target photograph is subjected to artificial intelligence scoring, the terminal device displays prompt information indicating a scoring process corresponding to the target photograph. And after the artificial intelligence scoring of the target photo is completed, displaying a quality evaluation result in a character expression form converted from the quality scoring of the target photo corresponding to the target photo. The terminal equipment can obtain the quality evaluation result of the character expression form corresponding to the target quality score. Or the terminal equipment combines the quality score with the preset quality evaluation character to obtain a quality evaluation result. For example, a quality score of 90 points is converted to "you beat 90% of people". Or convert the quality score of 20 points to "reprint photo bar" is not limited thereto.

According to the photo publishing method, when the target photo is subjected to artificial intelligence scoring, the prompt information in the scoring process is displayed corresponding to the target photo, so that a user can directly view the scoring progress; after the artificial intelligence scoring of the target photo is completed, the quality evaluation result in the character expression form converted from the quality scoring of the target photo is displayed corresponding to the target photo, so that the quality evaluation result can be displayed more intuitively, and a user can perform operations such as photo replacement according to the quality evaluation result.

In one embodiment, as shown in FIG. 2, a schematic interface diagram for displaying candidate photos in one embodiment is provided. The candidate photos are included in fig. 2: photograph A, photograph B, photograph C, photograph D, photograph E, photograph F, photograph G, photograph H, photograph I, photograph J, photograph K, photograph L, photograph M, photograph N, photograph O and photograph P. The target photos to be uploaded are arranged according to the selected sequence: photograph K, photograph J and photograph I. The interface of fig. 3 is displayed by a trigger operation on the "done" control of fig. 2. FIG. 3 is a schematic interface diagram of a photo confirmation page in one embodiment. In fig. 3, prompt information 302 indicating the scoring process and prompt information 304 indicating the scoring process are included. The representation of 302 is a progress bar and the representation of 304 is a text form. Fig. 3 also includes photograph K310, photograph J320, and photograph I330. Photograph K310, photograph J320, and photograph I330 are arranged in the order of selection of the interface in FIG. 2. Also included in fig. 3 is a photograph processing switch item 340 in the off state. In response to a triggering operation of the "save" control on the photo confirmation page, that is, a triggering of a posting operation, when the target photos are at least two, the target photos are posted in the order arranged in the photo confirmation page into the private page of the social application as in fig. 4. FIG. 4 is a diagram illustrating an interface of a personal page of a social application in one embodiment. In fig. 4, some of the photos, such as

photos

310 and 320, that the user has posted are shown. The personal page of the social application is posted in the order in which it was arranged in the photo confirmation page in fig. 4. Personal information of the user can also be displayed in fig. 4, for example, where the company is AAA company, profession (costume designer), graduate college (XX university), academic calendar (major), native place (guangdong Shenzhen), height (168cm), constellation (twin), hobbies (occasionally drinking | never smoking | do exercise).

FIG. 5 is a schematic interface diagram of a photo confirmation page in another embodiment. The quality evaluation results are displayed in the corresponding display areas of the target photographs in fig. 5, such as "beat 50% of people", "reprint photograph bar", and "beat 70% of people" in the figure. By triggering "re-transfer a few" as in fig. 5, more photos can be uploaded. Included in FIG. 5 is a photo processing switch item 350 in the on state. When the photo processing switch item is in an on state, in response to a 'save' operation triggered on the photo confirmation page, namely a posting operation, when the number of the target photos is at least two, the target photos sorted by the quality scores are posted to a personal page of the social application as shown in fig. 6. FIG. 6 is a schematic interface diagram of a personal page of a social application in another embodiment. Some of the user's published photos, such as photo 330 and photo 310, are shown in FIG. 6, and since photo 330 has a higher quality score than photo 310, photo 330 is ranked before photo 310. Personal information of the user can also be displayed in fig. 6, for example, where the company is AAA company, profession (costume designer), graduate college (XX university), academic calendar (major), native place (guangdong Shenzhen), height (168cm), constellation (twin), hobbies (occasionally drinking | never smoking | do exercise).

In an embodiment, as shown in fig. 7, a flowchart of an artificial intelligence based photo scoring method in an embodiment is applied to a social application scenario, and may be applied to a terminal device or a server, which is described by taking the application to the server as an example, and includes the following steps:

step 702, obtaining the target facial features of the target photo, and obtaining the target photo parameters of the target photo.

Wherein the facial features are used to represent features of a face. For example, the facial features may be at least one of, but not limited to, human facial features, cartoon characters facial features, and animal facial features. The facial features may include at least one of global facial features, distance facial features, area facial features, apparent facial features, texture facial features, skin condition features, hair style features, facial expression features. The target facial features refer to facial features of the target photograph. The picture parameters comprise at least one of resolution, brightness, sharpness, contrast, overall tone, size and position relation between a human body and the picture, shooting place and shooting time. The target photograph parameter refers to a photograph parameter of the target photograph.

Specifically, the server may extract target facial features of the target photograph through a facial recognition algorithm. The face recognition algorithm may be an ASM (Active Shape Models) algorithm, an HOG algorithm (Histogram of Oriented gradients), a convolutional neural network, or the like. And acquiring target picture parameters of the target picture.

In this embodiment, the reason why the shooting time and the shooting location are taken as the picture parameters is as follows: if the photos taken in the scenic spot are carefully taken by the user, the photos after five and six pm are generally underexposed from the shooting time, so the photos in the time period are ranked later.

And step 704, processing the target face features and the target photo parameters respectively by adopting at least two universal attraction models to obtain the similarity of each universal attraction model to the target photo. Each generic appeal model corresponds to a different base score.

The general attraction model also includes facial features and photo parameters. The general attraction model is constructed by facial features and photo parameters of the target photo existing in the social application. The face feature information and the photo parameter information corresponding to each universal attraction model are different. And each generic appeal model corresponds to a different base score. For example, the basic score corresponding to the general appeal model Z is 80 points, the basic score corresponding to the general appeal model Y is 60 points, the basic score corresponding to the general appeal model X is 40 points, and the basic score corresponding to the general appeal model W is 20 points. And the facial features and the photo parameters represented by the general attraction model Z, the general attraction model Y, the general attraction model X and the general attraction model W are different. The universal attraction model is trained according to the sample facial features, the sample photo parameters and the corresponding quality grading labels.

Specifically, the server adopts at least two general attraction models, and the target facial features and the target photo parameters are processed through each general attraction model to obtain the similarity of each general attraction model corresponding to the target photo. For example, the number of the general attraction models is 4, which are a general attraction model Z, a general attraction model Y, a general attraction model X, and a general attraction model W, respectively, then the general attraction model Z processes the target facial features and the target photo parameters, and the similarity of the general attraction model Z to the target photo is 90%; the universal attraction model Y processes the target facial features and the target photo parameters to obtain the similarity of the universal attraction model Y to the target photo, wherein the similarity of the universal attraction model Y to the target photo is 60%; the universal attraction model X processes the target facial features and the target photo parameters to obtain that the similarity of the universal attraction model X to the target photo is 50%; the universal attraction model W processes the target facial features and the target photo parameters to obtain the similarity of the universal attraction model W to the target photo of 30%.

And step 706, screening out a target general attraction model matched with the target photo according to the similarity.

Specifically, the target universal attraction model matched with the target photograph may refer to a universal attraction model having the highest similarity with the target photograph. For example, the general attraction model Z has a similarity of 90% to the target photograph, and the similarity is the highest, and then the general attraction model Z is the target general attraction model matching the target photograph.

And 708, generating a quality score of the target photo according to the basic score corresponding to the target general attraction model.

Wherein the quality score is used to characterize the attractiveness of the user's photo in the social application.

Specifically, the server generates a quality score of the target photo according to a basic score corresponding to the target general attraction model. The quality score of the target photograph may be the basic score, or may be a score obtained by combining the basic score with the similarity, but is not limited thereto.

According to the photo scoring method based on artificial intelligence, in social application, the attraction of a photo is determined not only by a color value, but also by a photo shooting technology and a later photo editing technology, so that the similarity of each universal attraction model to a target photo is obtained by processing based on target facial features and target photo parameters, the target universal attraction model matched with the target photo is screened out according to the similarity, and a quality score is generated based on basic score, namely the photo is scored by combining two kinds of information of the target facial features and the target photo parameters, the photo quality, namely the attraction of the whole photo in social software can be judged, and the accuracy of photo scoring can be improved.

In one embodiment, the target facial features include feature point facial features generated based on the facial feature points, and acquiring the target facial features of the target photo includes:

acquiring a target photo, and converting the target photo into a gray-scale image; carrying out face recognition based on the gray level image to obtain a face area in the target picture; extracting facial feature points from the facial region; and determining the feature points and the facial features of the target photo based on the facial feature points.

The grayscale map refers to an image represented by grayscale. The face region may be a region surrounded by a face contour. The feature point facial features are facial features that can be obtained based on the extracted facial feature points. For example, the feature point facial features include at least one of facial global features, facial distance features, and facial area features.

Specifically, the server acquires a target photo uploaded by the social application, converts the target photo into a gray-scale image, performs face recognition based on the gray-scale image to obtain a face area in the target photo, and extracts face feature points from the face area; and determining the feature points and the facial features of the target photo based on the facial feature points.

In this embodiment, the tilt angle of the face region may also be determined and corrected.

According to the artificial intelligence-based photo scoring method, the converted gray-scale image is subjected to face recognition to obtain a face region, the face feature points are extracted from the face region, and the feature point face features of the target photo are determined based on the face feature points.

In one embodiment, the feature point facial features include a facial global feature, a facial distance feature, and a facial area feature. Determining feature points and facial features of the target photo based on the facial feature points, comprising: combining and calculating the feature point coordinates corresponding to the face feature points to obtain face feature vectors, wherein the face feature vectors are used for representing face global features; determining the sizes of organs in the face and the distances among the organs based on the feature point coordinates to obtain face distance features; and determining the area surrounded by the facial feature points based on the feature point coordinates, determining the area of each area based on the feature point coordinates, and performing normalization processing on the area to obtain the facial area features.

Wherein the size of the organs in the face is the size of at least one of the five sense organs. For example, the facial distance feature may be at least one of a center-of-two-eye distance, a left-eye length, an inner-eye corner-of-two-eye distance, a right-eye length, a face width, a nose width, a face width across the tip of the nose, a face width across the upper lip, a length of the mouth, a face width across the lower lip, a width of the chin, an eyebrow and eye distance, an eye height, an eye-to-nose distance, a nose-to-mouth distance, a mouth-to-chin distance, an eyebrow-to-nose distance, a nose-to-chin distance. The area surrounded by the facial feature points may specifically be a triangular area, a quadrangular area, a pentagonal area, or the like, but is not limited thereto.

Specifically, feature point coordinates corresponding to the face feature points are combined and calculated, and a face feature vector for ensuring the global features of the face is obtained. For example, if there are 68 facial feature points, the coordinates of these 68 feature points are calculated according to a preset combination rule, and at least 34 facial feature vectors are obtained. The server determines the sizes of organs in the face and the distances among the organs based on the feature point coordinates, and obtains the face feature distances. And the server determines the area surrounded by the facial feature points based on the feature point coordinates according to a preset area calculation rule. The area is normalized, for example, to obtain 1 by normalizing the largest area, and the facial area features are obtained by dividing all other areas by the largest area.

In this embodiment, fig. 8 is a schematic diagram of a face distance feature in one embodiment. For example, fig. 8 includes the distance between the centers of the two eyes, the distance between the inner canthi of the two eyes, the length of the mouth, etc. FIG. 9 is a schematic diagram of a face distance feature in another embodiment. Included in fig. 9 are eye height, nose to mouth distance, mouth to chin distance, eyebrow and eye distance, etc. FIG. 10 is a schematic representation of facial area features in one embodiment. In fig. 10, the regions surrounded by the feature points are triangular regions, and the area of each triangular region may be determined based on the coordinates of the feature points, so as to obtain the facial area feature.

According to the artificial intelligence-based photo scoring method, the feature point coordinates corresponding to the face feature points are combined and calculated to obtain the face feature vector, so that the overall features of the face can be obtained; determining the sizes of organs in the face and the distances among the organs based on the feature point coordinates, so that the features of the organs of the face can be obtained; the method comprises the steps of determining areas surrounded by facial feature points based on feature point coordinates, determining the area of each area based on the feature point coordinates, and carrying out normalization processing on the areas, so that the distribution features of all organs of the face can be obtained, and scoring the photos by combining the features can improve the accuracy of scoring the photos in social application.

In one embodiment, generating a quality score of the target photo according to the base score corresponding to the target universal attraction model includes: acquiring the maximum additional score of the target general attraction model; determining the additional score of the target photo according to the similarity and the maximum additional score corresponding to the target general attraction model; and obtaining the quality score of the target photo according to the basic score and the additional score corresponding to the target general attraction model.

The maximum additional score of the target general attraction model is the difference value between the maximum score and the basic score corresponding to the target general attraction model. For example, the maximum score of the target universal attraction model is 100 points, the basic score is 80 points, and the maximum additional score is 100-80 points to 20 points.

Specifically, the server obtains the maximum additional score of the target general attraction model, and determines the additional score of the target photo according to the product of the similarity corresponding to the target general attraction model and the maximum additional score. And the server obtains the quality score of the target photo according to the sum of the basic score and the additional score corresponding to the target universal attraction model. For example, the target generic attraction model is the attraction model Z, corresponding to a base score of 80, a maximum additional score of 100-80-20, and a similarity of 90%, then an additional score of 90% × 20-18, and a quality score of 80+ 18-98.

According to the artificial intelligence-based photo scoring method, the additional score of the target photo is determined according to the similarity and the maximum additional score corresponding to the target general attraction model, the quality score of the target photo is obtained according to the basic score and the additional score corresponding to the target general attraction model, the similarity and the basic score can be combined, the obtained quality scoring result is more accurate, and a better photo quality suggestion can be provided for a user according to the result.

In one embodiment, as shown in FIG. 11, a flow diagram of an artificial intelligence photo scoring method in another embodiment is shown. And acquiring the target facial features of the target photo, acquiring the target photo parameters of the target photo, inputting the target facial features and the target photo parameters into the universal attraction model to obtain the quality score of the target photo, converting the quality score into a quality evaluation result according to the quality score, and displaying the quality evaluation result on the terminal equipment. If the similarity between the target facial features and the target photo parameters and the general attraction model A is high, the obtained quality score is related to the general attraction model A. And if the similarity between the target facial features and the target photo parameters and the general attraction model D is high, the obtained quality score is related to the general attraction model D.

In an embodiment, as shown in fig. 12, a flowchart of a training method of the universal attraction model in an embodiment is shown, where the universal attraction model is obtained by training through a training step, and the training step includes:

step 1202, at least two sample photos and quality score marks corresponding to the classified sample photos are obtained, and the types of the sample photos are used for representing different basic scores.

Wherein, the sample photo refers to a photo sample used for training the universal attraction model. Each sample photograph has a corresponding quality score label.

Specifically, the server obtains at least two sample photographs. Classifying the sample photos according to the basic scores, and acquiring quality score labels corresponding to the classified sample photos.

In this embodiment, the server may classify the sample photos based on the basic scores according to a preset number of untrained general attraction models, to obtain classified sample photos of classes in the preset number, where the preset number is at least two.

Step 1204, extract sample facial features and sample photograph parameters of the sample photograph.

Wherein the sample facial features are used to represent facial features of the sample photograph. For example, the sample facial features may be at least one of, but not limited to, sample facial features of a human, sample facial features of a cartoon character, and sample facial features of an animal. The sample facial features may include at least one of sample facial global features, sample facial distance features, sample facial area features, sample facial appearance features, sample facial texture features, sample skin condition features, sample hair style features, sample facial expression features. The sample photo parameters comprise at least one of resolution, brightness, sharpness, contrast, overall tone, size and position relation between human body and photo, shooting place and shooting time.

Specifically, the server extracts sample area features of the sample photograph and sample photograph parameters. The server may extract after classifying the sample photo, and may also extract the sample facial features of the sample photo and the sample photo parameters after obtaining the sample photo.

And 1206, processing by adopting at least two untrained universal attraction models respectively based on the sample facial features and the sample photo parameters of the corresponding classes to obtain the similarity of the untrained universal attraction models to the target photos of the corresponding classes, and corresponding basic scores of the untrained universal attraction models.

Specifically, at least two untrained general attraction models are adopted, each untrained general attraction model corresponds to the basic score, each untrained general attraction model processes the sample facial features and the sample parameters of the corresponding category, and the similarity of each untrained general attraction model to the target photo of the corresponding category is obtained. For example, the untrained generic attraction models are model Z, model Y, model X, and model W, corresponding to base scores of 80, 60, 40, and 20, respectively. The sample photos comprise a sample photo A, a sample photo B, a sample photo C and a sample photo D, wherein the score corresponding to A is 98 points, the score corresponding to B is 78 points, the score corresponding to C is 48 points, and the score corresponding to D is 38 points. The sample photograph of the model Z in the corresponding category is sample photograph a, the sample photograph of the model Y in the corresponding category is sample photograph B, the sample photograph of the model X in the corresponding category is sample photograph C, and the sample photograph of the model W in the corresponding category is sample photograph D.

And 1208, obtaining the sample quality score of the sample photo by combining the similarity according to the basic score corresponding to the untrained universal attraction model.

Specifically, the server obtains a sample quality score of the sample photo according to a basic score corresponding to the untrained universal attraction model and by combining the similarity.

In the embodiment, the maximum additional score of the target general attraction model is obtained; determining the additional score of the target photo according to the similarity and the maximum additional score corresponding to the target general attraction model; and obtaining the quality score of the target photo according to the basic score and the additional score corresponding to the target general attraction model.

And 1210, adjusting parameters in the untrained universal attraction model according to the sample quality score and the quality score label, and continuing training until the trained universal attraction model is obtained.

Specifically, the server adjusts parameters in the untrained universal attraction model according to the sample quality scores and the quality score labels, continues training until the iteration times or the loss value reaches the minimum value and the like, and obtains the trained universal attraction model.

According to the artificial intelligence-based photo scoring method, the sample facial features, the sample photo parameters and the quality scoring labels corresponding to the classified sample photos are used as training data of the universal attraction model, the untrained universal attraction model is trained, and a model capable of performing quality scoring on the photos of the corresponding category can be obtained through training; the sample facial features and sample photo parameters of the sample photo are extracted first, so that parameters needing to be trained in the universal attraction model can be reduced, and the training efficiency is improved.

In one embodiment, the obtaining of the quality score labels corresponding to the at least two sample photos and each classified sample photo includes: obtaining at least two sample photos and interaction data corresponding to each sample photo from a social application; classifying the sample photos based on the interaction data according to the preset number of untrained universal attraction models to obtain at least two classified sample photos; and carrying out normalization processing on each interactive data to obtain a quality score mark corresponding to each classified sample photo.

Wherein the interaction data is used to represent the evaluation of the photo posted by the first user by the second user in the social application. For example, the interaction data may be praise numbers or the like. The normalization process is a process of converting dimensional data into dimensionless data and converting the dimensionless data into a scalar quantity. For example, the normalization process may be to obtain data in the range of 0 to 1, or to obtain data in the range of 0 to 100.

Specifically, the server obtains at least two sample photos and interaction data corresponding to each sample photo from the social application. And classifying the sample photos by the server based on the interactive data according to the preset number of the untrained universal attraction models to obtain the classified sample photos. For example, if the number of untrained universal attraction models is 4 and the number of sample photos is 20, the sample photos may be arranged in the order of interaction data from high to low, and 5 photos are taken in the order as a class of photos, for a total of 4 classes of photos. And inputting the 4 types of photos into different universal attraction models respectively. And the server acquires the maximum value of the interactive data from the interactive data, and performs normalization processing on each piece of interactive data based on the maximum value of the interactive data to obtain the quality score mark corresponding to each classified sample photo. For example, if 2000 and 1000 are included in the interaction data, then normalizing 2000 and 1000 results in 100 and 50, respectively.

According to the photo scoring method based on artificial intelligence, at least two sample photos and interaction data corresponding to each sample photo are obtained from social application, classification is carried out based on the interaction data, normalization processing is carried out on the interaction data, quality scoring marks of the sample photos can be obtained by means of selection of most users in the social application, accordingly, a general attractive model which accords with public aesthetics is trained, quality scoring is not affected due to the will of a certain person, and the accuracy of the obtained photo quality scoring is improved.

In one embodiment, obtaining interaction data corresponding to each sample photo includes: for each sample photo, acquiring the corresponding praise number of each of at least two praise types; and carrying out weighting processing on the corresponding praise number based on the weight corresponding to each praise type to obtain interactive data.

Among them, the like may be "i like", "i like very much", "i like super like", etc., without being limited thereto. The weights corresponding to each of the at least two types of praise are not completely the same. For example, the weight corresponding to "i like" may be 0.25, the weight corresponding to "i like very much" may be 0.25, and the weight corresponding to "i super like" may be 0.5.

Specifically, for each sample photo, the server obtains the praise number corresponding to each of the at least two praise types, and performs weighting processing on the corresponding praise number based on the weight corresponding to each praise type to obtain the interactive data. For example, "like" corresponds to 1000 praise and a weight of 1; if the number of votes for "super like" is 100 and the weight is 3, then 1000 × 1+100 × 3 is 1300.

According to the artificial intelligence-based photo scoring method, the praise number corresponding to each of the praise types is obtained for each sample photo, the corresponding praise number is weighted based on the weight corresponding to each praise type, interaction data is obtained, the praise number can be expressed based on attractiveness of different degrees, and the interaction data is obtained through calculation, so that more accurate quality scoring marking is obtained, and the trained universal attraction model is more suitable for the public.

In one embodiment, as shown in fig. 13, a flow diagram of a training mode of the universal attraction model in another embodiment is shown. And extracting the characteristics of the sample photo, and extracting the sample facial characteristics and the sample photo parameters of the sample photo. Acquiring interaction data corresponding to each sample photo, classifying the sample photos based on the interaction data according to the preset number of untrained universal attraction models to obtain classified sample photos, and performing normalization processing on each interaction data to obtain quality score marks corresponding to each classified sample photo. And performing machine learning on the sample facial features, the sample photo parameters and the quality score labels to obtain sample quality scores, adjusting parameters in the untrained universal attraction model according to the sample quality scores and the quality score labels, and continuing training until a trained universal attraction model is obtained. Machine Learning (ML) is a multi-domain cross discipline, and relates to a plurality of disciplines such as probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and the like. The special research on how a computer simulates or realizes the learning behavior of human beings so as to acquire new knowledge or skills and reorganize the existing knowledge structure to continuously improve the performance of the computer. Machine learning is the core of artificial intelligence, is the fundamental approach for computers to have intelligence, and is applied to all fields of artificial intelligence. Machine learning and deep learning generally include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and formal education learning.

In one embodiment, the target facial features include at least one of global facial features, facial distance features, facial area features, facial appearance features, facial texture features, skin condition features, hair style features, facial expression features; the target picture parameter comprises at least one of resolution, brightness, sharpness, contrast, overall tone, size and position relation between a human body and the picture, shooting place and shooting time.

The facial appearance features refer to the overall appearance of the face, facial skin features, facial line features and other features. The facial skin features may be, for example, whether there is acne, whether the skin is dark, whether the skin is depressed, etc. Facial texture features refer to features in the face such as wrinkles. Facial expression features such as smiling, crying, anger, and other emotional features. The resolution is directly extracted from the information carried by the target photo. Luminance the average luminance of a photograph can be calculated by converting the photograph into HSL (Hue, Saturation, luminance) color space. The contrast of the target picture reflects the sharpness of the target picture to a certain extent, wherein one calculation method is to sum the squares of the differences between the gray value of the central pixel and the gray values of the surrounding 4 neighboring pixels and then divide the sum by the number of the differences.

The code is as follows:

wherein img is a target photo, i and j respectively represent the abscissa and the ordinate of a pixel point, m and n are respectively the maximum of the abscissa and the maximum of the ordinate of the photo, and cg is the contrast.

The calculation of the color tone is an average value of RGB (Red, Green, Blue, Red, Green and Blue) components of the integral pixel point, and the optimized algorithm can be used for uniformly sampling and obtaining a color average value.

In the artificial intelligence-based photo scoring method, the target facial features include at least one of global facial features, facial distance features, facial area features, facial appearance features, hair style features and facial expression features; the parameters of the target photo comprise at least one of resolution, brightness, sharpness, contrast, integral tone, size and position relation between a human body and the photo, shooting location and shooting time, and the target photo can be scored by combining features of different dimensions, namely the target photo is scored based on public aesthetics, so that the scoring accuracy is improved.

In one embodiment, as shown in FIG. 14, a flowchart of an artificial intelligence based photo scoring method in yet another embodiment is shown. Target photograph parameters and target facial features of the target photograph are obtained. The target facial parameters include at least one of resolution, brightness, sharpness, contrast, overall hue, and human-to-photograph size relationship. The target facial feature extraction comprises the following steps of preprocessing the picture: and acquiring a target photo, and converting the target photo into a gray-scale image. And performing face recognition based on the gray-scale image to obtain a face area in the target picture. Facial feature points are extracted from the facial region. And determining the global facial feature, the facial distance feature and the facial area feature of the target photo based on the facial feature points. And processing the at least two general attraction models respectively based on the target facial features and the target photo parameters to obtain the similarity of each general attraction model to the target photo, wherein each general attraction model corresponds to different basic scores. And screening out a target universal attraction model matched with the target photo according to the similarity. And generating a quality score of the target photo according to the basic score corresponding to the target general attraction model. And converting the quality scores into corresponding quality score results. And in response to a publishing operation triggered by the photo confirmation page, publishing the target photos sorted according to the quality scores to a personal page of the social application when the number of the target photos is at least two.

In one embodiment, an artificial intelligence based photo scoring method further comprises: displaying a quality evaluation result of the target photograph corresponding to the target photograph; the quality evaluation result represents a quality score obtained by carrying out artificial intelligence scoring on the target photo; and in response to a publishing operation triggered by the photo confirmation page, publishing the target photos sorted according to the quality scores to a personal page of the social application when the number of the target photos is at least two. According to the artificial intelligence-based photo scoring method, information such as facial features, photo parameters, quality evaluation results and praise numbers in social application are integrated, universal attraction models of different levels are established, after a user uploads photos, the user is helped to extract the facial features, the attraction of the photos of the user is judged according to the similarity between the facial features and the universal attraction models, and the user is advised to replace the photos with low attraction; if the user uploads a plurality of photos, different face models in the photos are respectively established, and then according to the face similarity, the attractive force of one of the photos is judged to be higher, so that the photos of the user are intelligently helped to carry out intelligent recommendation sequencing.

In one embodiment, obtaining target facial features of a target photograph includes: when at least two face regions are included in the target picture, determining the integrity of organs in each face region; when the integrity of each face region is the same, extracting the target face features in the face region with the largest area in the face regions; and when the integrity degrees are different, extracting the target facial features corresponding to the facial region with the highest integrity degree in the facial regions. The method can also comprise the following steps: and when the face areas corresponding to the face areas are the same, determining the definition of each face area, and acquiring the target face characteristics corresponding to the target area with the highest definition of the face area.

According to the artificial intelligence-based photo scoring method, when one photo comprises at least two facial regions, the integrity of organs is determined firstly, and the highest integrity is selected; the integrity is the same, and the area is the largest; the areas are the same, the main angle with the highest definition can be determined in the target photo, feature extraction is carried out on the main angle, scoring is carried out, and accuracy of photo scoring is improved.

In one embodiment, obtaining target facial features of a target photograph includes: acquiring reference facial features with the largest occurrence frequency in published photos in a user account; determining a face region in the target photograph; when at least two face regions are included in the target photograph, the face feature having the highest similarity with the reference face feature is taken as the target face feature. According to the artificial intelligence-based photo scoring method, when the target photo comprises at least two facial regions, the face with the largest number of occurrences in the published photo is generally the face of the user, and the face with the highest similarity to the face features with the largest number of occurrences in the published photo is more likely to be the face of the user, so that the face of the belonging user can be evaluated.

In one embodiment, a method of training a generic attraction model includes: acquiring at least two sample photos and quality score marks corresponding to the classified sample photos, wherein the types of the sample photos are used for representing different basic scores; extracting sample facial features and sample photo parameters of the sample photo; processing the samples based on the facial features and the photo parameters of the samples in the corresponding categories respectively by adopting at least two untrained universal attraction models to obtain the similarity of the untrained universal attraction models to the target photos in the corresponding categories; the untrained universal attraction model corresponds to the basic score; obtaining a sample quality score of the sample photo according to a basic score corresponding to the untrained universal attraction model and by combining the similarity; and adjusting parameters in the untrained universal attraction model according to the sample quality score and the quality score label, and continuing training until the trained universal attraction model is obtained.

It should be understood that although the various steps in the flowcharts of fig. 1, 7 and 12 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least some of the steps in fig. 1, 7 and 12 may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, which are not necessarily performed in sequence, but may be performed in turn or alternately with other steps or at least some of the other steps or stages.

In one embodiment, as shown in fig. 15, there is provided a photo publishing apparatus, which may be a part of a computer device using a software module or a hardware module, or a combination of the two, the apparatus specifically includes: an upload module 1502, a display module 1504, and a publish module 1506, wherein:

an upload module 1502, configured to display the candidate photos in response to a photo upload operation triggered in the social application, and obtain target photos to be uploaded in response to a selection operation on the candidate photos;

a display module 1504, configured to enter a photo confirmation page and display a target photo in a state of responding to an editing operation;

a display module 1504, configured to display a quality evaluation result of the target photo corresponding to the target photo; the quality evaluation result represents a quality score obtained by carrying out artificial intelligence scoring on the target photo;

and the publishing module 1506 is used for responding to a publishing operation triggered on the photo confirmation page, and publishing the target photos sorted according to the quality scores to a private page of the social application when the number of the target photos is at least two.

The photo publishing device carries out artificial intelligent grading on the target photo uploaded by the user to obtain a quality grading result and display the quality grading result, and can objectively reflect the quality of the target photo uploaded by the user; in response to the publishing operation triggered by the photo confirmation page, when the number of the target photos is at least two, the target photos sequenced according to the quality scores are published to the personal page of the social application, that is, the target photos sequenced according to the quality scores can be displayed on the personal page of the social application, the position of the photos does not need to be manually adjusted by a user, the photo publishing efficiency is improved, the user can obtain more attention in the social application, and the user experience is improved.

In one embodiment, the display module 1504 is configured to display a photo processing switch item on a photo confirmation page; responding to the triggering operation of the photo processing switch item, and adjusting the state of the photo processing switch item to be one of an opening state and a closing state;

the publishing module 1506 is configured to, when the photo processing switch item is in an on state, respond to a publishing operation triggered on the photo confirmation page, and publish the target photos sorted according to the quality scores to a private page of the social application when the target photos are at least two.

According to the photo publishing device, the photo confirmation page comprises the photo processing switch item, when the photo processing switch item is in an open state, the target photos sequenced according to the quality scores are published to the personal page of the social application, and then a user can independently select whether intelligent sequencing is needed or not, and when the photo processing switch item is selected to be opened, the user does not need to manually adjust the positions of the photos, so that the photo publishing efficiency is improved.

In one embodiment, the target photos displayed in the photo confirmation page are arranged in the selected order; the publishing module 1506 is configured to, in response to a publishing operation triggered on the photo confirmation page when the photo processing switch item is in the off state, publish the target photos to the private page of the social application in an order arranged in the photo confirmation page when the target photos are at least two.

According to the photo publishing device, when the photo processing switch items are in the closed state, under the condition that the number of the target photos is at least two, the target photos are published in the personal page applied by the social contact according to the sequence arranged in the photo confirmation page, and a user can independently select whether intelligent sequencing is needed or not, so that the user experience is improved.

In one embodiment, the posting module 1506 is configured to display the highest quality-scored target photograph on the cover page of the personalized page in response to the posting operation triggered on the photo confirmation page when the photo handling switch item is in the on state.

According to the photo publishing device, when the photo processing switch item is in the on state, the target photo with the highest quality score is displayed on the cover of the personal page in response to the publishing operation triggered by the photo confirmation page, so that the attraction of the user in the social application can be improved, and more attention can be obtained.

In one embodiment, the editing operation comprises a replacement operation. The uploading module 1502 is configured to re-display the candidate photos in response to a replacement operation triggered on the target photo displayed in the photo confirmation page, and acquire the target photo to be replaced in response to a selection operation on the re-displayed photo; when the picture confirmation page is returned, replacing the target picture acted by the replacement operation with the target picture to be replaced; the display module 1504 is used for displaying the quality evaluation result of the replaced target photo corresponding to the replaced target photo.

The photo publishing device responds to the replacement photo triggered by the target photo, replaces the target photo acted by the replacement operation with the target photo to be replaced, still displays the quality evaluation result of the replaced target photo, and can replace the photo with poor quality when the quality evaluation result is poor, so that the quality of the published photo is improved.

In one embodiment, the display module 1504 is configured to display prompt information representing a scoring process corresponding to a target photo when performing artificial intelligence scoring on the target photo;

and after the artificial intelligence scoring of the target photo is completed, displaying a quality evaluation result in a character expression form converted from the quality scoring of the target photo corresponding to the target photo.

When the photo publishing device is used for carrying out artificial intelligent scoring on the target photo, the prompting information in the scoring process is displayed corresponding to the target photo, so that a user can directly view the scoring progress; after the artificial intelligence scoring of the target photo is completed, the quality evaluation result in the character expression form converted from the quality scoring of the target photo is displayed corresponding to the target photo, so that the quality evaluation result can be displayed more intuitively, and a user can perform operations such as photo replacement according to the quality evaluation result.

In one embodiment, as shown in fig. 16, there is provided a photo publishing apparatus, which may be a part of a computer device using a software module or a hardware module, or a combination of the two, the apparatus specifically includes: a feature obtaining module 1602, a similarity obtaining and screening module 1606, and a quality score generating module 1608, wherein:

a feature obtaining module 1602, configured to obtain a target facial feature of the target photo, and obtain a target photo parameter of the target photo;

a similarity obtaining module 1604, configured to use at least two general attraction models, and perform processing based on the target facial features and the target photo parameters, respectively, to obtain a similarity of each general attraction model to the target photo; each universal attraction model corresponds to a different basic score;

a screening module 1606, configured to screen out a target universal attraction model matching the target photo according to the similarity;

the quality score generation module 1608 is configured to generate a quality score of the target photo according to the basic score corresponding to the target general attraction model.

According to the photo scoring device based on artificial intelligence, in social application, the attraction of a photo is determined not only by a color value, but also by a photo shooting technology and a later photo editing technology, so that the similarity of each universal attraction model to the target photo is obtained by processing based on target facial features and target photo parameters, the target universal attraction model matched with the target photo is screened out according to the similarity, and a quality score is generated based on basic score, namely the photo is scored by combining two kinds of information of the target facial features and the target photo parameters, the photo quality, namely the attraction of the whole photo in social software can be judged, and the accuracy of photo scoring can be improved.

In one embodiment, the feature obtaining module 1602 is configured to obtain a target photo, and convert the target photo into a gray-scale image; carrying out face recognition based on the gray level image to obtain a face area in the target picture; extracting facial feature points from the facial region; and determining the feature points and the facial features of the target photo based on the facial feature points.

According to the artificial intelligence-based photo scoring device, the converted gray-scale image is subjected to face recognition to obtain a face region, the face feature points are extracted from the face region, and the feature point face features of the target photo are determined based on the face feature points.

In one embodiment, the feature point facial features include a facial global feature, a facial distance feature, and a facial area feature;

the feature obtaining module 1602 is configured to perform combination calculation on feature point coordinates corresponding to the face feature points to obtain a face feature vector, where the face feature vector is used to represent global features of a face;

and determining the area surrounded by the facial feature points based on the feature point coordinates, determining the area of each area based on the feature point coordinates, and performing normalization processing on the area to obtain the facial area features.

The artificial intelligence-based photo scoring device performs combined calculation on the feature point coordinates corresponding to the face feature points to obtain face feature vectors, so that the overall features of the face can be obtained; determining the sizes of organs in the face and the distances among the organs based on the feature point coordinates, so that the features of the organs of the face can be obtained; the method comprises the steps of determining areas surrounded by facial feature points based on feature point coordinates, determining the area of each area based on the feature point coordinates, and carrying out normalization processing on the areas, so that the distribution features of all organs of the face can be obtained, and scoring the photos by combining the features can improve the accuracy of scoring the photos in social application.

In one embodiment, the quality score generation module 1608 is for obtaining a maximum additional score for the target generic attraction model;

determining the additional score of the target photo according to the similarity and the maximum additional score corresponding to the target general attraction model;

and obtaining the quality score of the target photo according to the basic score and the additional score corresponding to the target general attraction model.

According to the photo scoring device based on artificial intelligence, the additional score of the target photo is determined according to the similarity and the maximum additional score corresponding to the target general attraction model, the quality score of the target photo is obtained according to the basic score and the additional score corresponding to the target general attraction model, the similarity and the basic score can be combined, the obtained quality scoring result is more accurate, and a better photo quality suggestion can be given to a user according to the result.

In one embodiment, the artificial intelligence-based photo scoring device comprises a training module, wherein the training module is used for acquiring at least two sample photos and quality scoring labels corresponding to the classified sample photos, and the categories of the sample photos are used for representing different basic scores; extracting sample facial features and sample photo parameters of the sample photo; processing the samples based on the facial features and the photo parameters of the samples in the corresponding categories respectively by adopting at least two untrained universal attraction models to obtain the similarity of the untrained universal attraction models to the target photos in the corresponding categories; the untrained universal attraction model corresponds to the basic score; obtaining a sample quality score of the sample photo according to a basic score corresponding to the untrained universal attraction model and by combining the similarity; and adjusting parameters in the untrained universal attraction model according to the sample quality score and the quality score label, and continuing training until the trained universal attraction model is obtained.

The artificial intelligence-based photo scoring device adopts the sample facial features, the sample photo parameters and the quality scoring labels corresponding to the classified sample photos as training data of the universal attraction model, trains the untrained universal attraction model, and can train to obtain a model capable of scoring the quality of the photos of the corresponding category; the sample facial features and sample photo parameters of the sample photo are extracted first, so that parameters needing to be trained in the universal attraction model can be reduced, and the training efficiency is improved.

In one embodiment, the training module is used for acquiring at least two sample photos and interaction data corresponding to each sample photo from a social application; classifying the sample photos based on the interaction data according to the preset number of untrained universal attraction models to obtain at least two classified sample photos; and carrying out normalization processing on each interactive data to obtain a quality score mark corresponding to each classified sample photo.

According to the photo scoring device based on artificial intelligence, at least two sample photos and interactive data corresponding to each sample photo are obtained from social application, classification is carried out on the interactive data, normalization processing is carried out on the interactive data, quality scoring marks of the sample photos can be obtained by means of selection of most users in the social application, accordingly, a general attractive model which accords with public aesthetics is trained, quality scoring is not affected due to the will of a certain person, and the accuracy of the obtained photo quality scoring is improved.

In one embodiment, the training module is configured to, for each sample photo, obtain the like number corresponding to each of the at least two like types; and carrying out weighting processing on the corresponding praise number based on the weight corresponding to each praise type to obtain interactive data.

According to the artificial intelligence-based photo scoring device, the praise number corresponding to each of the praise types in the at least two praise types is obtained for each sample photo, the corresponding praise number is weighted based on the weight corresponding to each praise type, interaction data is obtained, the praise number can be expressed based on different degrees of attraction, and the interaction data is obtained through calculation, so that more accurate quality scoring marking is obtained, and the universal attraction model obtained through training is more suitable for the public.

In one embodiment, the target facial features include at least one of global facial features, facial distance features, facial area features, facial appearance features, hair style features, facial expression features; the target picture parameter comprises at least one of resolution, brightness, sharpness, contrast, overall tone, size and position relation between a human body and the picture, shooting place and shooting time.

According to the artificial intelligence-based photo scoring device, the target facial features comprise at least one of global facial features, facial distance features, facial area features, facial appearance features, hair style features and facial expression features; the parameters of the target photo comprise at least one of resolution, brightness, sharpness, contrast, integral tone, size and position relation between a human body and the photo, shooting location and shooting time, and the target photo can be scored by combining features of different dimensions, namely the target photo is scored based on public aesthetics, so that the scoring accuracy is improved.

For the specific definition of the photo publishing device, reference may be made to the above definition of the photo publishing method, and for the specific definition of the artificial intelligence based photo scoring device, reference may be made to the above definition of the artificial intelligence based photo scoring method, and details of the description about the artificial intelligence based photo scoring method are omitted here. All or part of the modules in the photo publishing device and the artificial intelligence based photo evaluation device can be realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.

In one embodiment, a computer device is provided, and the computer device may be a terminal device, and its internal structure diagram may be as shown in fig. 17. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used for carrying out wired or wireless communication with an external terminal, and the wireless communication can be realized through WIFI, an operator network, NFC (near field communication) or other technologies. The computer program is executed by a processor to implement a photo distribution method. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, a key, a track ball or a touch pad arranged on the shell of the computer equipment, an external keyboard, a touch pad or a mouse and the like.

Those skilled in the art will appreciate that the architecture shown in fig. 17 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

In one embodiment, a computer device is further provided, which includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of the above method embodiments when executing the computer program.

In an embodiment, a computer-readable storage medium is provided, in which a computer program is stored which, when being executed by a processor, carries out the steps of the above-mentioned method embodiments.

In one embodiment, a computer program product or computer program is provided that includes computer instructions stored in a computer-readable storage medium. The computer instructions are read by a processor of a computer device from a computer-readable storage medium, and the computer instructions are executed by the processor to cause the computer device to perform the steps in the above-mentioned method embodiments.

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database or other medium used in the embodiments provided herein can include at least one of non-volatile and volatile memory. Non-volatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical storage, or the like. Volatile Memory can include Random Access Memory (RAM) or external cache Memory. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others.

The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.

The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.

Claims

1. A method for publishing a photo, the method comprising:

responding to a photo uploading operation triggered in the social application, displaying a candidate photo, responding to a selection operation of the candidate photo, and acquiring a target photo to be uploaded;

2. The method of claim 1, further comprising:

displaying a photo processing switch item on the photo confirmation page;

and when the photo processing switch item is in an on state, executing the step of responding to the publishing operation triggered by the photo confirmation page, and publishing the target photos sequenced according to the quality scores to the personal page of the social application when the number of the target photos is at least two.

3. The method of claim 2, wherein the target photos displayed in the photo confirmation page are arranged in a selected order; the method further comprises the following steps:

and when the photo processing switch item is in a closed state, responding to a publishing operation triggered on the photo confirmation page, and publishing the target photos to the personal page of the social application according to the sequence arranged in the photo confirmation page when the target photos are at least two.

4. A method according to claim 2 or 3, characterized in that the method further comprises:

and when the photo processing switch item is in an on state, responding to the issuing operation triggered on the photo confirmation page, and displaying the target photo with the highest quality score on the cover of the personal page.

5. The method of claim 1, wherein the editing operation comprises a replacement operation, the method further comprising:

responding to a replacing operation triggered by the target photo displayed in the photo confirmation page, displaying the candidate photo again, responding to a selecting operation of the photo displayed again, and acquiring the target photo to be replaced;

when the photo confirmation page is returned, replacing the target photo acted by the replacement operation with the target photo to be replaced;

6. The method according to claim 1, 2, 3 or 5, wherein the displaying the quality evaluation result of the target photo corresponding to the target photo comprises:

when the target photo is subjected to artificial intelligence scoring, displaying prompt information representing a scoring process corresponding to the target photo;

7. An artificial intelligence based photo scoring method, the method comprising:

8. The method of claim 7, wherein the target facial features comprise feature point facial features generated based on facial feature points, and wherein obtaining the target facial features of the target photograph comprises:

acquiring a target photo, and converting the target photo into a gray-scale image;

carrying out face recognition based on the gray level image to obtain a face area in the target picture;

extracting facial feature points from the facial region;

and determining feature points and facial features of the target photo based on the facial feature points.

9. The method of claim 8, wherein the feature point facial features comprise a facial global feature, a facial distance feature, and a facial area feature;

the determining feature points and facial features of the target photo based on the facial feature points comprises:

combining and calculating the feature point coordinates corresponding to the facial feature points to obtain facial feature vectors, wherein the facial feature vectors are used for representing the global facial features;

10. The method of claim 7, wherein generating the quality score of the target photo according to the base score corresponding to the target universal attraction model comprises:

acquiring the maximum additional score of the target universal attraction model;

11. The method of claim 7, wherein the generic attraction model is trained by a training step comprising:

obtaining at least two sample photos and quality score marks corresponding to the classified sample photos, wherein the types of the sample photos are used for representing different basic scores;

12. The method of claim 11, wherein the obtaining of the at least two sample photographs and the quality score label corresponding to each of the classified sample photographs comprises:

obtaining at least two sample photos and interaction data corresponding to each sample photo from a social application;

13. The method of claim 12, wherein obtaining interaction data corresponding to each sample photograph comprises:

for each sample photo, acquiring the corresponding praise number of each of at least two praise types;

14. A photograph posting apparatus, characterized in that the apparatus comprises:

15. An artificial intelligence based photograph scoring apparatus, the apparatus comprising: