Image propagation method, device, equipment and computer program product

Abstract

The invention provides an image propagation method, apparatus and device, and a computer program product. The method comprises the following steps: determining identity information of an object user in a target image; and based on the identity information of the object user, acquiring a privacy requirement opinion for spreading the target image from the object user. And performing privacy protection processing matched with the privacy requirement opinion on the area of the object user in the target image. And spreading the target image subjected to the privacy protection processing. The problem that privacy of others is exposed when images are propagated arbitrarily can be solved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an image transmission method, device, equipment and computer program product. Background Art

[0002] Thanks to the rapid development of mobile Internet and the popularity of smart phones around the world, social networks have also ushered in a golden period of development. With the dissemination of a large amount of information every day, the privacy and security issues of social networks have gradually become prominent. In particular, when users post image data such as photos and videos on social networks, they often intentionally or unintentionally reveal the private information of others, which brings great security risks to others.

[0003] Therefore, how to prevent users from spreading images without authorization and exposing the privacy of others is an issue that needs to be solved urgently. Summary of the invention

[0004] The purpose of this application is to provide an image dissemination method, device, equipment and computer program product, which can solve the problem of exposing other people's privacy when users disseminate images without authorization.

[0005] In order to achieve the above purpose, the embodiment of the present application is implemented as follows: In a first aspect, an image transmission method is provided, comprising: Determine the identity information of the target user in the target image; Based on the identity information of the target user, obtaining a privacy requirement opinion for disseminating the target image from the target user; performing a privacy protection process matching the privacy requirement opinion on the area of ​​the target user in the target image; The target image after the privacy protection process is performed is propagated.

[0006] In a second aspect, an image transmission device is provided, comprising: A determination module, used to determine the identity information of the target user in the target image; An acquisition module, configured to acquire, from the target user, a privacy requirement opinion for disseminating the target image based on the identity information of the target user; An execution module, configured to execute a privacy protection process matching the privacy requirement opinion on the area of ​​the target user in the target image; The dissemination module is used to send an authorization request for disseminating the target image to the target user when the privacy requirement prediction result of the target user meets the preset requirement.

[0007] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor; and a memory configured to store computer-executable instructions, wherein the computer-executable instructions, when executed, cause the processor to execute the method described in the first aspect.

[0008] According to a fourth aspect, a computer program product is provided, the computer program product comprising a computer-readable storage medium storing a computer program, wherein the computer program is operable to cause a computer to execute the method according to the first aspect.

[0009] In the embodiment of the present application, before the target image needs to be disseminated, the identity information of the target user in the target image is first determined, and then the privacy requirements for disseminating the target image are actively solicited from the target user based on the identity information. The target image is then disseminated after corresponding privacy protection processing is performed in accordance with the privacy requirements fed back by the target user, thereby avoiding the problem of exposing the privacy of others by disseminating images without their consent. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0011] Figure 1 A schematic diagram of the flow of the image propagation method according to an embodiment of the present application.

[0012] Figure 2 A schematic diagram of the process of training a deep model for the image propagation method according to an embodiment of the present application.

[0013] Figure 3 It is a schematic diagram of the structure of the image transmission device according to an embodiment of the present application.

[0014] Figure 4 A schematic diagram of the structure of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0015] As mentioned above, as an information sharing platform, social networks inevitably have privacy and security risks. In particular, when users spread image data such as photos and videos, they often intentionally or unintentionally reveal other people's private information, which brings great security risks to others.

[0016] According to statistics, the dissemination of group photos is the most vulnerable to exposing other people's privacy. Group photos involve the portrait privacy of multiple users. Some users are more sensitive to privacy, while others are more casual about it. If group photos are directly posted on social networks, it will cause disgust among more sensitive users. Therefore, it is necessary to desensitize the users in the group photos by retouching them before posting.

[0017] Obviously, it is not convenient to retouch the images for the target users one by one. Therefore, this application aims to propose an intelligent image dissemination solution, which can determine the identity information of the target users in the image before disseminating the image, and actively obtain the privacy requirements of the disseminated image from the target users, so as to perform privacy protection processing on the image according to the privacy requirements before dissemination, so as to avoid the problem of exposing the privacy of others without consent.

[0018] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0019] An embodiment of the present application provides an image propagation method. Figure 2 is a flowchart of the image propagation method, including: S102, determining the identity information of the target user in the target image.

[0020] In this embodiment, the target image is an image containing at least one target user screen to be disseminated in a social network. The social network described here may be, but is not limited to, a chat group, social media, etc. In addition, the dissemination behavior may be one-to-one dissemination or one-to-many dissemination, which is not specifically limited in this article.

[0021] Specifically, this embodiment can obtain the identity information of the target user in the target image from the disseminator of the target image, and / or determine the identity information of the target user in the target image based on face recognition technology.

[0022] Taking obtaining the identity information of the target user from the communicator as an example, the communicator can actively mark the identity information of each target user in the target image before submitting the target image. The marking method can be manually setting an electronic tag for indicating the identity information in the area of ​​each target user in the target image.

[0023] Taking the example of determining the identity information of the target user based on face recognition technology, an identity recognition model based on facial features can be pre-configured. After the communicator submits the target image, the target image is input into the identity recognition model, and the identity recognition model identifies the corresponding identity information based on the facial features of each target user in the target image.

[0024] It should be noted that there is not only one way to confirm identity information, and this article does not make any specific restrictions here.

[0025] S104, based on the identity information of the target user, obtaining the privacy requirement opinion of the target image to be disseminated from the target user.

[0026] In this embodiment, the privacy requirement opinion reflects whether the target user needs privacy protection or does not need privacy protection.

[0027] In order to improve the efficiency of soliciting privacy requirements, with the support of big data, the historical personal records of the target user can be first obtained based on the identity information of the target user. The historical personal records need to contain information showing the degree of privacy emphasis of the target user. Afterwards, the privacy requirements of the target user are predicted based on the historical personal records to obtain the predicted results of the privacy requirements of the target user. If the predicted results of the privacy requirements of the target user meet the preset privacy requirements standard (high privacy requirements), the privacy requirements opinions of the target user for propagating the target image are obtained. On the contrary, if the predicted results of the privacy requirements of the target user do not meet the preset privacy requirements standard (low privacy requirements), the privacy requirements opinions of the target user can be selected.

[0028] As an example, historical personal records may include information on dimensions such as information publishing behavior, personal portrait, privacy protection settings, behavioral preferences, and interactive behavior.

[0029] Among them, information publishing behavior may include: at least one of personal image publishing frequency and personal image access rights; personal portrait may include: at least one of age group, gender and occupation; privacy protection settings may include: at least one of privacy setting level, privacy setting frequency and password security strength; behavioral preferences may include: at least one of published image type distribution and privacy information filtering ratio; interactive behavior may include: at least one of interactive frequency and interactive information type.

[0030] Taking information publishing behavior as an example: if the target user publishes personal images frequently, it means that he often uploads his personal selfies on social networks and does not mind his facial information being disclosed on social networks. Therefore, the corresponding level of privacy attention is relatively low.

[0031] Taking personal portraits as an example: users of young age, highly sensitive occupations, and highly sensitive topic genders attach greater importance to privacy.

[0032] Taking the privacy protection setting as an example: if the password security strength set by the target user is high and / or the privacy setting level is high (non-friends cannot see the friend circle, and stranger information is rejected), the corresponding privacy emphasis is high; otherwise, the corresponding privacy emphasis is low.

[0033] Take behavioral preference as an example: if the distribution of image types posted by the target user indicates that the proportion of his / her personal selfies posted is much higher than the proportion of his / her personal selfies sent, it means that he / she does not like to share personal photos on social networks, and therefore would mind others uploading their photos to social networks, and therefore attaches great importance to privacy.

[0034] Taking interactive behavior as an example: if the target user often shares photos and chats with other account users, the corresponding level of privacy emphasis is low; otherwise, the corresponding level of privacy emphasis is high.

[0035] Obviously, the above historical personal records can reflect the degree of privacy emphasis of the target user and can be used as a basis for predicting privacy requirements.

[0036] In practical applications, this embodiment can use artificial intelligence technology to predict the privacy requirements of the target user based on the historical personal records of the target user.

[0037] As an exemplary introduction: First, feature extraction is performed on historical personal records based on at least two feature dimensions of information release behavior, personal portrait, privacy protection settings, behavioral preferences, and interactive behavior to obtain feature data of the target user corresponding to at least two feature dimensions; then, nonlinear transformation and fusion are performed on the feature data of the target user corresponding to at least two feature dimensions to obtain cross-feature data of the target user; then, the cross-feature data of the target user is input into a deep learning model to predict the privacy requirements of the target user by the deep learning model.

[0038] The ability of the deep learning model to predict privacy requirements is obtained by training based on the cross-feature data of the sample object users and the corresponding privacy requirement labels. The privacy requirement labels are used to mark the privacy requirements of the sample object users. The cross-feature data of the sample object users is obtained by nonlinearly transforming the feature data of at least two feature dimensions of the sample object users. The training principle is as follows Figure 2As shown in the figure: the cross-feature data of the sample object user is input into the deep learning model to predict the privacy requirements of the sample object user and obtain the predicted results of the privacy requirements of the sample object user; then, the privacy requirements of the sample object user indicated by the privacy requirement label are taken as the true value, the error loss between the predicted results of the privacy requirements of the sample object user and the privacy requirement label is calculated, and the model parameters of the deep learning model are adjusted for the purpose of reducing the error loss. In this way, the model parameters of the deep learning model are continuously iterated until the changes in the model parameters converge, so that the predicted results of the privacy requirements of the sample object users provided by the deep learning model can be close to the privacy requirements indicated by the privacy requirement label as the true value, thereby obtaining the ability to predict privacy requirements based on cross-feature data.

[0039] In addition to determining the privacy importance of the target user through the historical personal records of the target user, this embodiment can also perform image semantic analysis on the target image to determine whether the target user in the target image needs privacy protection based on the image semantic analysis result.

[0040] Among them, image semantics refers to the meaning of image content, which can be expressed through natural language.

[0041] For example, if the target image is a photo taken when the target user participated in XXX outdoor activities, the image semantics may be “XXX outdoor activity photo”.

[0042] For another example, if the target image is a photo of the target user taken at a family dinner, the image semantics may be “family dinner photo”.

[0043] It should be understood that image semantics can reflect the privacy level of image content to a certain extent. Using the above example, generally, "outdoor activity photos" belong to public content and are mostly used for external sharing, so the corresponding privacy level is relatively low, and it can be determined that the target user does not need privacy protection; while "family gathering photos" belong to non-public content and are mostly used for internal sharing, so the corresponding privacy level is relatively high, and it is necessary to actively ask the target user whether to protect privacy.

[0044] As a feasible implementation method, the present embodiment can recognize the image semantics of the target image through a large language model. The method is to set the prompt word of the large language model to "You play the role of image content recognition and determine the image semantics expressed by the input image", and then input the target image into the large language model, and the large language model can recognize the image semantics of the target image according to the instructions of the prompt word.

[0045] It should be noted that recognizing image semantics is an existing capability of the large language model. This embodiment uses the image semantic recognition results provided by the large language model to determine whether the target image is non-public content. Based on the foregoing, if the target image is non-public content, it indicates a high degree of privacy. Therefore, it is necessary to further obtain the privacy requirements for disseminating the target image from the target user based on the identity information of the target user in the target image.

[0046] S106: Performing a privacy protection process that matches the privacy requirement on the area of ​​the target user in the target image.

[0047] Specifically, this embodiment can perform privacy protection processing on the area of ​​the target user in the target image when the privacy requirement opinion indicates that the target user needs privacy protection. Among them, the privacy protection processing can be but is not limited to blurring, blocking, replacing and cropping, etc., which are not specifically limited in this article. Preferably, in order to minimize the impact of photo editing on the target image, a suitable method can be selected for privacy protection processing according to the position of the area of ​​the target user in the target image. For example: if the area of ​​the target user is at the edge of the target image, the area of ​​the target user in the target image is cropped, that is, the size of the target image display area is modified to directly eliminate the target user; if the area of ​​the target user is not at the edge of the target image, cropping is no longer appropriate. In this case, at least one of blocking, blurring and replacement can be used to process the area of ​​the target user in the target image.

[0048] In addition, if the privacy requirement opinion indicates that the target user does not need privacy protection, this embodiment may not modify the target user area in the target image. It should be noted that not modifying the target user area in the target image is also a privacy protection processing method.

[0049] S108, disseminating the target image after the privacy protection processing is performed.

[0050] In summary, based on the method of the embodiment, before the target image needs to be disseminated, the identity information of the target user in the target image is first determined, and then the information presenting the privacy emphasis of the target user is obtained based on the identity information to predict the privacy requirements of the target user. If the privacy requirements of the target user meet certain standards, it is necessary to solicit the privacy requirements of the target user for the dissemination of the target image to understand whether the target user minds. For the target user who minds, the target image is disseminated after the privacy protection processing is performed on the picture area of ​​the target image, so as to avoid the problem of exposing the privacy of others without consent. In related applications, the method of the embodiment can intelligently analyze the target users who attach a high degree of importance to privacy in the image, and actively solicit privacy requirements for privacy protection processing, which brings great convenience compared to the communicator performing it himself.

[0051] The following describes the method flow of this embodiment in combination with actual application scenarios.

[0052] In this application scenario, the communicator publishes the group photo on social media or a chat group. The method of this embodiment is executed by an application of the social media or chat group, and the corresponding steps include: Step 1: Obtain the group photo submitted by the communicator to be published.

[0053] In this step, the communicator can select the group photo to be published from the photo album of the personal terminal through the application of social media or chat group.

[0054] Step 2: Determine the identity information of each user in the group photo.

[0055] As mentioned above, in this embodiment, the identity information of each object user in the group photo can be determined by face recognition or the communicator's own labeling, which will not be repeated here.

[0056] Specifically, the communicator can actively mark the users in the group photo who need to be authorized through the application. The application only needs to determine the identity information of the marked users, and choose to ignore the unmarked users in the group photo, and will no longer seek authorization from the unmarked users in the future.

[0057] In addition, if there are target users whose identity information has not been identified, the application can choose to refuse to post the photo on social media or chat groups, which will not be elaborated here.

[0058] Step three: Based on the deep learning model, predict the privacy requirements of the subject users in the group photo.

[0059] It should be understood that different target users have different privacy requirements, and this application scenario can be analyzed from the historical personal records left by these target users in social media or chat groups.

[0060] Correspondingly, this step extracts features from the historical personal records of each object user based on at least two feature dimensions of information publishing behavior, personal portrait, privacy protection settings, behavioral preferences, and interactive behavior, and obtains feature data corresponding to at least two feature dimensions for each object user.

[0061] As an exemplary introduction, the description of feature data can refer to the following table:

[0062] Afterwards, the feature data of at least two feature dimensions corresponding to each object user are nonlinearly transformed and fused to obtain cross-feature data of each object user, and the cross-feature data of each object user is input into the deep learning model to predict the privacy requirements of each object user.

[0063] In this application scenario, the privacy requirements predicted by the deep learning model are reflected in the form of scores. The higher the score, the higher the privacy requirement, and the lower the score, the lower the privacy requirement. Correspondingly, the training process of the deep learning model is as follows: 1) Prepare a training set for the deep learning model, which includes the historical personal records of the sample object user and the corresponding score of the sample object user. The score of the sample object user is the privacy requirement label described in the text.

[0064] 2) Based on at least two feature dimensions of information release behavior, personal portrait, privacy protection settings, behavioral preferences, and interactive behavior, feature extraction is performed on the historical personal records of the sample object user to obtain feature data corresponding to at least two feature dimensions of the sample object user.

[0065] 3) Perform nonlinear transformation and fusion on the feature data of at least two feature dimensions of the sample object user to obtain cross-feature data of the sample object user.

[0066] 4) Input the cross-feature data of the sample object users into the deep learning model and try to predict the score of the sample object users.

[0067] 5) The loss function is used to calculate the error loss between the score predicted by the deep learning model and the score annotated with the privacy requirement label.

[0068] 6) Back-propagate and adjust the model parameters of the deep learning model to reduce error loss.

[0069] The values ​​of the privacy requirement label and the privacy requirement prediction result are represented by one-hot vectors. One-hot encoding is used to convert discrete classification labels into binary vectors. In One-Hot encoding, a unique binary bit is assigned to each classification variable, and the bit is used to represent the value of the variable. If the variable value is 1, the corresponding binary bit is 1; if the value is 0, the corresponding binary bit is 0. In One-Hot encoding, each variable is encoded only once, so it is called "one bit effective".

[0070] For example, suppose there is a categorical variable with three categories: A, B, and C. Using One-Hot encoding, one binary bit can be assigned to each of the three categories. If a sample belongs to category A, the binary bit representing A in its One-Hot vector is 1, while the binary bits representing B and C are both 0; if a sample belongs to category B, the binary bit representing B in its One-Hot vector is 1, while the binary bits representing A and C are both 0; and so on.

[0071] As an example introduction: Taking the 5-point privacy requirement score as an example, during the training process, the feature data and privacy requirement labels of each feature dimension can be encoded using one_hot encoding.

[0072] Each value of the one-hot vector corresponds to a binary bit. For example, the privacy requirement label with a score of 3 is encoded as [0,0,1,0,0] according to one_hot encoding; for another example, the privacy requirement label with a score of 4 is encoded as [0,0,0,4,0] according to one_hot encoding; correspondingly, the encoding process is expressed as:

[0073] in, is the encoding result of the privacy requirement label, is the scoring integer value of the input privacy requirement, and one_hot is the encoding function.

[0074] Correspondingly, the deep learning model prediction process is expressed as:

[0075] in, ' represents the output of the deep learning model training phase; , and Represents the input, weight matrix, and bias of a deep learning model; represents the learning rate; represents the loss function; represents the activation function; .

[0076] Since the scores output by the deep learning model are expressed in one-hot encoding, they need to be converted into integer scores, namely:

[0077] in, Indicates One_hot encoding output corresponding to training samples; is the corresponding integer privacy sensitivity score; It is the inverse process of the one_hot encoding function.

[0078] The corresponding loss function uses the mean square error function:

[0079] in, Indicates the number of training samples; Indicates The score given by the privacy requirement label corresponding to the training sample (i.e., the value of the privacy requirement label); Indicates that the deep learning model is given The score of the training sample prediction (i.e., the value of the privacy-required prediction result); Represents error loss.

[0080] Step 4: Solicit the opinions of target users whose scores reach the preset threshold on the privacy requirements for publishing group photos.

[0081] Still taking the 5-point privacy requirement rating as an example, in this step, the application can solicit privacy requirement opinions on the release of group photos from target users who have reached 4 points.

[0082] Since the target user is also a user of the application, in this step, the application can send the photo and a text message of whether privacy protection is needed to the account of the target user who has reached 4 points through the system information window, and receive the privacy requirements feedback from the target user.

[0083] Step 5: For users whose privacy requirements indicate that privacy protection is required, any method such as blurring, blocking, replacing and cropping is selected to perform privacy protection processing on the area in the group photo.

[0084] For example, if the user whose privacy needs to be protected is in the edge area of ​​the group photo, he / she will be directly removed from the photo by cropping; if the user whose privacy needs to be protected is in the middle area of ​​the group photo, he / she will be blocked or blurred.

[0085] In summary, in this application scenario, when the communicator initiates the dissemination of a group photo on social media or a chat group, the social media or chat group application identifies the identity information of each user in the group photo, and searches for the corresponding historical personal records based on the identity information to predict the privacy requirements; for the user with higher privacy requirements, the privacy requirements for the group photo are actively sought, and the image area in the group photo is processed for privacy protection according to the privacy requirements, and then the group photo is published on social media or chat groups according to the communicator's operation. In the whole process, identity recognition, privacy requirement prediction, privacy requirement opinion solicitation, and privacy protection processing can be performed by the application, and the communicator cannot bypass the application to publish the group photo without authorization, so it can effectively curb the problem of the communicator intentionally or unintentionally exposing the privacy of others on social media or chat groups.

[0086] Corresponds to Figure 1 The method shown, another embodiment of the present application provides an image transmission device. Figure 3 3 is a schematic diagram of the structure of the image transmission device 300, comprising: The determination module 310 is used to determine the identity information of the target user in the target image.

[0087] The acquisition module 320 is used to obtain the privacy requirement opinion of the target image from the target user based on the identity information of the target user.

[0088] The execution module 330 is used to perform a privacy protection process matching the privacy requirement opinion on the area of ​​the target user in the target image.

[0089] The propagation module 340 is configured to send an authorization request for propagating the target image to the target user if the predicted result of the privacy requirement of the target user meets a preset requirement.

[0090] Before the target image needs to be disseminated, the device of this embodiment first determines the identity information of the target user in the target image, and then actively solicits the privacy requirements of the target image from the target user based on the identity information, and then disseminates the target image after performing corresponding privacy protection processing according to the privacy requirements fed back by the target user, thereby avoiding the problem of exposing the privacy of others by disseminating images without their consent.

[0091] Optionally, the acquisition module 320 obtains the privacy requirement opinion for disseminating the target image from the target user based on the identity information of the target user, including: obtaining the historical personal record of the target user based on the identity information; the historical personal record contains information presenting the degree of privacy emphasis of the target user; predicting the privacy requirement of the target user based on the historical personal record to obtain the predicted result of the privacy requirement of the target user; and obtaining the privacy requirement opinion for disseminating the target image from the target user when the predicted result of the privacy requirement of the target user meets the preset privacy requirement standard.

[0092] Optionally, the acquisition module 320 predicts the privacy requirements of the target user based on the historical personal records, including: extracting features from the historical personal records based on at least two feature dimensions of information publishing behavior, personal portrait, privacy protection settings, behavioral preferences, and interactive behavior to obtain feature data of the target user corresponding to the at least two feature dimensions; performing nonlinear transformation and fusion on the feature data of the target user corresponding to the at least two feature dimensions to obtain cross-feature data of the target user; and inputting the cross-feature data of the target user into a deep learning model to predict the privacy requirements of the target user.

[0093] Optionally, the image transmission device of this embodiment includes: A training module is used to train the deep learning model based on the cross-feature data of sample object users and the corresponding privacy requirement labels before inputting the cross-feature data of the object user into the deep learning model to predict the privacy requirements of the object user; wherein the cross-feature data of the sample object user is obtained by performing a nonlinear transformation on the feature data of the sample object user corresponding to the at least two feature dimensions, and the privacy requirement label is used to mark the privacy requirements of the sample object user.

[0094] Optionally, the training module trains the deep learning model based on the cross-feature data of the sample object users and the corresponding privacy requirement labels, including: inputting the cross-feature data of the sample object users into the deep learning model to predict the privacy requirements of the sample object users and obtain the privacy requirement prediction results of the sample object users; calculating the error loss between the privacy requirement prediction results of the sample object users and the privacy requirement labels, and adjusting the model parameters of the deep learning model with the purpose of reducing the error loss.

[0095] Optionally, the calculation formula of the error loss is: ; in, represents the error loss, represents the number of training samples, represents the value of the privacy requirement label of the mth sample object user, Represents the value of the privacy requirement prediction result of the mth sample object user.

[0096] Optionally, the training module adjusts the model parameters of the deep learning model for the purpose of reducing the error loss, including: based on the formula , adjusting the model parameters of the deep learning model; in, , represents the input of the deep learning model, represents the output of the deep learning model, represents the model parameters of the deep learning model, Indicates deviation, represents the learning rate, represents the loss function, Represents the activation function.

[0097] Optionally, the values ​​of the privacy requirement label and the privacy requirement prediction result are represented by a one-hot vector, and each value of the one-hot vector corresponds to a binary bit.

[0098] Optionally, the acquisition module 320 obtains the privacy requirement opinions for disseminating the target image from the target user based on the identity information of the target user, including: performing image semantic recognition on the target image based on a large language model to obtain an image semantic recognition result corresponding to the target image; judging whether the target image is non-public content based on the image semantic recognition result; if the target image is non-public content, obtaining the privacy requirement opinions for disseminating the target image from the target user based on the identity information of the target user.

[0099] Optionally, the execution module 330 performs privacy protection processing on the area of ​​the object user in the target image that matches the privacy requirement opinion, including: when the privacy requirement opinion indicates that the object user needs privacy protection, performing privacy protection processing of at least one of blurring, blocking, replacing and cropping on the area of ​​the object user in the target image.

[0100] Optionally, the information publishing behavior includes: at least one of personal image publishing frequency and personal image access rights; The personal portrait includes: at least one of age group, gender and occupation; The privacy protection setting includes: at least one of a privacy setting level, a privacy setting frequency, and a password security strength; The behavior preference includes: at least one of the distribution of published image types and the proportion of privacy information filtering; The interactive behavior includes: at least one of an interactive frequency and an interactive information type.

[0101] It should be noted that the image transmission of this embodiment can be used as Figure 1 The execution subject of the method shown can thus realize Figure 1 The steps and functions in the method shown are not repeated here.

[0102] Figure 4 This is a schematic diagram of the structure of an electronic device provided by an embodiment of the present application. Figure 4 At the hardware level, the electronic device includes a processor, and optionally also includes an internal bus, a network interface, and a memory. The memory may include a memory, such as a high-speed random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage. Of course, the electronic device may also include hardware required for other services.

[0103] The processor, network interface and memory can be interconnected through an internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0104] The memory is used to store the program. Specifically, the program may include a program code, and the program code includes a computer operation instruction. The memory may include a memory and a non-volatile memory, and provides instructions and data to the processor.

[0105] The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it, forming the above-mentioned image transmission device at the logical level. The processor executes the program stored in the memory and is specifically used to perform the following operations: Determine the identity information of the target user in the target image.

[0106] Based on the identity information of the target user, a privacy requirement opinion for disseminating the target image is obtained from the target user.

[0107] A privacy protection process matching the privacy requirement opinion is performed on the area of ​​the target user in the target image.

[0108] The target image after the privacy protection process is performed is propagated.

[0109] The above application Figure 1 The method disclosed in the illustrated embodiment can be applied to a processor or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in the processor or an instruction in the form of software. The above processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in one or more embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in combination with one or more embodiments of the present application can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module can be located in a storage medium mature in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0110] Of course, in addition to software implementation methods, the electronic device of the present application does not exclude other implementation methods, such as logic devices or a combination of software and hardware, etc. That is to say, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0111] The present application also provides a computer program product, which includes a computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the following operations: Determine the identity information of the target user in the target image.

[0112] Based on the identity information of the target user, a privacy requirement opinion for disseminating the target image is obtained from the target user.

[0113] A privacy protection process matching the privacy requirement opinion is performed on the area of ​​the target user in the target image.

[0114] The target image after the privacy protection process is performed is propagated.

[0115] In short, the above description is only a preferred embodiment of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of the present application shall be included in the protection scope of one or more embodiments of the present application.

[0116] The systems, devices, and modules described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0117] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0118] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0119] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

Claims

1. An image propagation method, characterized in that: include: Determine the identity information of the target user in the target image; Based on the identity information of the target user, obtaining a privacy requirement opinion for disseminating the target image from the target user; performing a privacy protection process matching the privacy requirement opinion on the area of ​​the target user in the target image; The target image after the privacy protection process is performed is propagated.

2. The method according to claim 1, characterized in that Based on the identity information of the target user, obtaining the privacy requirement opinion of the target user for disseminating the target image, including: Acquire the historical personal record of the target user based on the identity information; wherein the historical personal record includes information showing the degree of privacy emphasis of the target user; Predicting the privacy requirement of the target user based on the historical personal records to obtain a prediction result of the privacy requirement of the target user; When the predicted result of the privacy requirement of the target user meets the preset privacy requirement standard, the privacy requirement opinion of the target image is obtained from the target user.

3. The method according to claim 2, characterized in that Predicting the privacy requirements of the target user based on the historical personal records includes: Based on at least two feature dimensions of information publishing behavior, personal portrait, privacy protection settings, behavior preferences, and interactive behavior, feature extraction is performed on the historical personal records to obtain feature data of the target user corresponding to the at least two feature dimensions; Performing nonlinear transformation and fusion on the feature data of the target user corresponding to the at least two feature dimensions to obtain cross-feature data of the target user; The cross-feature data of the target user is input into a deep learning model to predict the privacy requirement of the target user.

4. The method according to claim 3, characterized in that Before inputting the cross-feature data of the target user into the deep learning model to predict the privacy requirement of the target user, the method further includes: The deep learning model is trained based on the cross-feature data of the sample object users and the corresponding privacy requirement labels; wherein the cross-feature data of the sample object users are obtained by performing a nonlinear transformation on the feature data of the sample object users corresponding to the at least two feature dimensions, and the privacy requirement labels are used to mark the privacy requirements of the sample object users.

5. The method according to claim 4, characterized in that The deep learning model is trained based on the cross-feature data of the sample object users and the corresponding privacy requirement labels, including: Inputting the cross-feature data of the sample object users into the deep learning model to predict the privacy requirements of the sample object users, and obtaining the privacy requirement prediction results of the sample object users; The error loss between the privacy requirement prediction result of the sample object user and the privacy requirement label is calculated, and the model parameters of the deep learning model are adjusted for the purpose of reducing the error loss.

6. The method according to claim 5, characterized in that The calculation formula of the error loss is: ; in, represents the error loss, represents the number of training samples, represents the value of the privacy requirement label of the mth sample object user, Represents the value of the privacy requirement prediction result of the mth sample object user.

7. The method according to claim 5, characterized in that Adjusting the model parameters of the deep learning model for the purpose of reducing the error loss includes: Based on the formula , adjusting the model parameters of the deep learning model; in, , represents the input of the deep learning model, represents the output of the deep learning model, represents the model parameters of the deep learning model, Indicates deviation, represents the learning rate, represents the loss function, Represents the activation function.

8. The method according to claim 7, characterized in that The values ​​of the privacy requirement label and the privacy requirement prediction result are represented by a one-hot vector, and each value of the one-hot vector corresponds to a binary bit.

9. The method according to claim 1, characterized in that: Based on the identity information of the target user, obtaining the privacy requirement opinion of the target user for disseminating the target image, including: Based on the large language model, performing image semantic recognition on the target image to obtain an image semantic recognition result corresponding to the target image; Based on the image semantic recognition result, determining whether the target image is non-public content; If the target image is non-public content, based on the identity information of the target user, a privacy requirement opinion for disseminating the target image is obtained from the target user.

10. The method according to any one of claims 1 to 9, characterized in that: Performing a privacy protection process matching the privacy requirement opinion on the area of ​​the target user in the target image, comprising: In a case where the privacy requirement opinion indicates that the target user requires privacy protection, at least one of blurring, blocking, replacing and cropping is performed on the region of the target user in the target image.

11. The method according to claim 10, characterized in that At least one of blurring, blocking, replacing and cropping the area of ​​the object user in the target image comprises: If the area of ​​the target user in the target image belongs to the edge of the target image, the area of ​​the target user in the target image is cropped; otherwise, the area of ​​the target user in the target image is at least one of blocked, blurred and replaced.

12. The method according to any one of claims 3 to 8, characterized in that The information publishing behavior includes: at least one of personal image publishing frequency and personal image access rights; The personal portrait includes: at least one of age group, gender and occupation; The privacy protection setting includes: at least one of a privacy setting level, a privacy setting frequency, and a password security strength; The behavior preference includes: at least one of the distribution of published image types and the proportion of privacy information filtering; The interactive behavior includes: at least one of an interactive frequency and an interactive information type.

13. The method according to any one of claims 1 to 8, characterized in that Determine the identity information of the object user in the target image, including: Obtaining identity information of a target user in the target image from a disseminator of the target image; And / or, determining the identity information of the object user in the target image based on face recognition technology.

14. The method according to any one of claims 1 to 8, characterized in that The target image after the privacy protection process is transmitted, comprising: The target image after the privacy protection process is performed is published to a designated chat group.

15. An image transmission device, characterized in that: include: A determination module, used to determine the identity information of the target user in the target image; An acquisition module, configured to acquire, from the target user, a privacy requirement opinion for disseminating the target image based on the identity information of the target user; An execution module, configured to execute a privacy protection process matching the privacy requirement opinion on the area of ​​the target user in the target image; The dissemination module is used to send an authorization request for disseminating the target image to the target user when the privacy requirement prediction result of the target user meets the preset requirement.

16. An electronic device comprising: processor; and a memory arranged to store computer executable instructions which, when executed, cause the processor to perform a method as claimed in any one of claims 1 to 14.

17. A computer program product, comprising a computer-readable storage medium storing a computer program, wherein the computer program is operable to cause a computer to execute the method according to any one of claims 1 to 14.

Images