A method, apparatus, computer device, and storage medium for sending images.

Abstract

This disclosure provides a method, apparatus, computer device, and storage medium for sending images. The method includes: receiving an image to be reviewed sent by a client; the image to be reviewed is an image that the client determines does not match a stored problematic image after comparing it with the image to be sent; searching for a target image matching the image to be reviewed from a target image library based on the signature information of the image to be reviewed; responding to the finding of a matching target image, determining whether the image to be reviewed is a problematic image based on the problem level of the target images stored in the target image library; and if the image to be reviewed is not a problematic image, sending it to other clients. This disclosure, through a two-level review process, can effectively detect whether an image to be sent by a client is a problematic image, thereby improving the accuracy of problematic image detection.

Description

Technical Field

[0001] This disclosure relates to the field of image processing technology, and more specifically, to a method, apparatus, computer device, and storage medium for sending images. Background Technology

[0002] With the widespread use of the internet, people are increasingly relying on it to spread information such as images. For example, in instant messaging (IM) scenarios, people can use instant messaging tools to send pictures to friends.

[0003] To ensure that submitted images meet requirements, backend staff typically need to review them to check for illegal, sensitive, or other problematic images. The number of images awaiting review is usually enormous, and relying solely on manual review can easily result in missed problematic images. Summary of the Invention

[0004] This disclosure provides at least one method, apparatus, computer device, and storage medium for sending images.

[0005] In a first aspect, embodiments of this disclosure provide a method for sending images, including:

[0006] Receive images to be reviewed sent by the client; the images to be reviewed are those that the client determines do not match the problem images after comparing the images to be sent with the problem images stored in the client.

[0007] Based on the signature information of the image to be reviewed, a target image matching the image to be reviewed is searched from the target image library;

[0008] Upon finding a target image that matches the image to be reviewed, the system determines whether the image to be reviewed is a problematic image based on the problem level of the target image stored in the target image library.

[0009] If it is determined that the image to be reviewed is not a problematic image, the image to be reviewed will be sent to other clients.

[0010] In one feasible implementation, after searching for a target image that matches the image to be reviewed from the target image library, the method further includes:

[0011] If no matching target image is found, the image to be reviewed is sent to another client, and a pre-trained problem identification model is used to extract image features from the image to be reviewed. Based on the extracted image features, the problem level of the image to be reviewed is determined; and / or,

[0012] If no matching target image is found, the image to be reviewed is sent to another client, and the text information in the image to be reviewed is extracted. The extracted text information is compared with preset sensitive text to obtain the comparison result, and the problem level of the image to be reviewed is determined based on the comparison result.

[0013] The images to be reviewed and the determined issue levels are saved to the target image library.

[0014] In one feasible implementation, the method further includes:

[0015] Obtain interaction details information for each image in the target image library; the interaction details information includes at least one of the following: number of interactions, number of participants in the interaction group, number of saves, and number of favorites;

[0016] Based on the interaction details of each image, the interaction popularity of each image is determined;

[0017] For each image, if the interaction popularity of the image reaches a set threshold, the problem level of the image is re-examined, and the problem level after re-examination is determined; the re-examination includes manual review and / or review using more models than the initial review.

[0018] The re-examined issue level is saved to the target image library.

[0019] In one feasible implementation, the method further includes:

[0020] Obtain the interaction count and issue level of each image in the target image library;

[0021] For each image, if the number of interactions with the image meets a first preset condition and the problem level of the image meets a second preset condition, the image is determined to be a problem image. Upon receiving a problem image update request from the client, the problem image is sent to the client, and the problem image is used by the client to update the stored problem images.

[0022] Secondly, this disclosure also provides a method for sending images, including:

[0023] Get the image to be sent;

[0024] The image to be sent is compared with the first problem image stored locally to determine whether the image to be sent matches the first problem image;

[0025] If the image to be sent does not match the first problematic image, the image to be sent to the server for further review.

[0026] In one feasible implementation, the method further includes:

[0027] Send a request to the server to update the problematic image;

[0028] Receive the second problem image sent by the server in response to the problem image update request;

[0029] If the sum of the number of the second problem image and the first problem image exceeds a set threshold, a target problem image to be deleted is determined based on the duration information of when the first problem image was not matched, and the target problem image is deleted.

[0030] In one feasible implementation, determining the target problematic image to be deleted based on the duration information of the first problematic image not being matched includes:

[0031] The longest unmatched first question image among the first question images is identified as a candidate question image to be deleted;

[0032] Determine whether the sum of the number of remaining first question images (excluding the candidate question images) and the number of second question images is less than the set threshold;

[0033] If the threshold is less than the threshold, then based on the interaction popularity and / or issue level of the candidate issue images, a target issue image to be deleted is determined from the candidate issue images, until the sum of the number of remaining first issue images and second issue images equals the set threshold after deleting the target issue image.

[0034] In one feasible implementation, after determining whether the sum of the number of remaining first problem images (excluding the candidate problem images) and the number of second problem images is less than the set threshold, the method further includes:

[0035] If the value is greater than the threshold, after deleting the candidate problem image, the step of determining the longest unmatched first problem image among the remaining first problem images as the candidate problem image to be deleted continues until the sum of the number of remaining first problem images and the number of second problem images equals the set threshold.

[0036] In one feasible implementation, comparing the image to be sent with a first problem image stored locally to determine whether the image to be sent matches the first problem image includes:

[0037] Generate signature information for the image to be sent, compare the hash value of the signature information of the image to be sent with the hash value of the signature information of the first problem image stored locally, and determine whether the image to be sent matches the first problem image.

[0038] Thirdly, embodiments of this disclosure also provide an image sending device, comprising:

[0039] The receiving module is used to receive images to be reviewed sent by the client; the images to be reviewed are images that the client determines do not match the problem images after comparing the images to be sent with the problem images stored in the client.

[0040] The search module is used to search for a target image that matches the image to be reviewed from the target image library based on the signature information of the image to be reviewed.

[0041] The first determining module is used to respond to the discovery of a target image that matches the image to be reviewed, and to determine whether the image to be reviewed is a problematic image based on the problem level of the target image stored in the target image library;

[0042] The sending module is used to send the image to be reviewed to other clients if it is determined that the image to be reviewed is not a problematic image.

[0043] Fourthly, embodiments of this disclosure also provide an image sending device, comprising:

[0044] The acquisition module is used to acquire images to be sent.

[0045] The determination module is used to compare the image to be sent with the first problem image stored locally to determine whether the image to be sent matches the first problem image;

[0046] The first sending module is used to send the image to be sent to the server for re-review if the image to be sent does not match the first problem image.

[0047] Fifthly, embodiments of this disclosure also provide a computer device, including: a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the computer device is running, the processor communicates with the memory via the bus, and when the machine-readable instructions are executed by the processor, the steps of the first aspect above, or any possible implementation of the first aspect, or the steps of the second aspect above, or any possible implementation of the second aspect, are executed.

[0048] In a sixth aspect, embodiments of this disclosure also provide a computer-readable storage medium storing a computer program that, when executed by a processor, performs the steps of the first aspect or any possible implementation of the first aspect, or performs the steps of the second aspect or any possible implementation of the second aspect.

[0049] In the image sending method provided in this embodiment, the image to be reviewed is the image that the client determines is a non-problem image after comparing the image to be sent with the problem images stored on the client. In other words, the image to be reviewed, which is a non-problem image after the client has completed local review, is reviewed again on the server. This two-level review method can reduce the burden on the client's local storage on the one hand, and effectively detect whether the image to be sent by the client is a problem image on the other hand, thereby improving the accuracy of problem image detection.

[0050] To make the above-mentioned objects, features and advantages of this disclosure more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings used in the embodiments will be briefly described below. These drawings are incorporated in and constitute a part of this specification. They illustrate embodiments conforming to this disclosure and, together with the specification, serve to explain the technical solutions of this disclosure. It should be understood that the following drawings only show some embodiments of this disclosure and should not be considered as limiting the scope. Those skilled in the art can obtain other related drawings based on these drawings without creative effort.

[0052] Figure 1 A flowchart of an image sending method provided by an embodiment of this disclosure is shown;

[0053] Figure 2 A flowchart of another image sending method provided by an embodiment of this disclosure is shown;

[0054] Figure 3 A schematic diagram of an image sending device provided in an embodiment of this disclosure is shown;

[0055] Figure 4 A schematic diagram of another image sending device provided in an embodiment of this disclosure is shown;

[0056] Figure 5 A schematic diagram of a computer device provided in an embodiment of this disclosure is shown;

[0057] Figure 6A schematic diagram of another computer device provided in an embodiment of this disclosure is shown. Detailed Implementation

[0058] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. The components of the embodiments of this disclosure described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this disclosure provided in the accompanying drawings is not intended to limit the scope of the claimed disclosure, but merely represents selected embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.

[0059] In instant messaging scenarios, people can send pictures to friends using instant messaging tools. Some pictures may involve illegal or sensitive content. The presence of such inappropriate pictures is detrimental to the healthy development of the online environment. Therefore, to ensure that sent pictures meet requirements, backend personnel usually need to review them to check whether they are illegal or sensitive. When the number of pictures to be reviewed is very large, relying solely on manual review can easily result in missed violations.

[0060] Based on this, this disclosure provides an image sending method. After receiving an image to be reviewed from a client, the method searches for a matching target image in a target image library based on the signature information of the image to be reviewed. Upon finding a matching target image, the method determines whether the image to be reviewed is a problematic image based on the problem level of the target images stored in the target image library. If the image to be reviewed is determined not to be a problematic image, it is then sent to other clients. In this image sending process, images that did not match the problem image criteria after local review by the client are reviewed again on the server. This two-level review reduces the burden on local client storage and effectively detects whether images to be sent by the client are problematic, thereby improving the accuracy of problematic image detection.

[0061] The deficiencies of the above solutions and the proposed solutions are the result of the inventor's practice and careful research. Therefore, the discovery process of the above problems and the solutions proposed in this disclosure below should be considered as the inventor's contribution to this disclosure.

[0062] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0063] To facilitate understanding of this embodiment, a method for sending images disclosed in this disclosure will first be described in detail. The execution subject of the image sending method provided in this disclosure is generally a computer device with certain computing power.

[0064] The image sending method disclosed in this embodiment is mainly applied in scenarios where image information can be sent, such as instant messaging. The instant messaging client sends an image to the server, which then reviews the image. If the review is successful, the server can send the approved image to other instant messaging clients; if the review fails, the server cannot send the failed image to other instant messaging clients.

[0065] The image sending method provided in this embodiment will be described below, taking the server as the executing entity as an example.

[0066] See Figure 1 The diagram shows a flowchart of an image sending method provided in this embodiment of the present disclosure. The method includes steps S101 to S104, wherein:

[0067] S101: Receive an image to be reviewed sent by the client; the image to be reviewed is an image that the client determines does not match the problem image after comparing the image to be sent with the problem image stored in the client.

[0068] In this embodiment, the image to be reviewed can refer to an image that has already been reviewed on the client side and needs to be reviewed again on the server side. The image to be reviewed can be an image that did not match any of the problem images stored on the client side during the client-side review. That is, when comparing the image to be reviewed with the problem images stored on the client side, no matching problem image was found. Here, a problem image can refer to an image whose interaction count meets a first preset condition and whose problem level meets a second preset condition. The problem level can be pre-set based on the content contained in the image; the image content can be the reason for the problem in the image, and different content can correspond to different problem levels. The correspondence between image content and problem level can be pre-set and is not specifically limited here. The process of determining whether an image is a problem image will be described below.

[0069] When an image to be reviewed is processed on the client side, one method involves comparing the hash value of the signature information of the image to be reviewed with the hash value of the signature information of the problematic image stored on the client side, and determining whether a problematic image was not matched based on the comparison result. The signature information can be MD5 data generated using a Message-Digest Algorithm (MD5). To reduce image data size and save data storage space, another method involves cropping the image to be reviewed before processing, obtaining a cropped image, and then using the aforementioned MD5 algorithm to generate the corresponding MD5 data for the cropped image. The image to be reviewed can be cropped to a preset size according to specific cropping requirements; no specific limitation is required here.

[0070] The problem image stored on the client can be sent by the server to the client. In specific implementations, the problem image can be sent proactively by the server; it can also be retrieved by the client from the server, for example, the client can retrieve it from the server via a short connection; or it can be sent by the server in response to a problem image update request sent by the client. There are no specific limitations here. To ensure timely updates of problem images while reducing data transmission pressure on both the server and client, in one approach, the problem image stored on the client can be sent periodically, such as once per hour.

[0071] In this embodiment of the disclosure, in addition to the method described above where the server directly sends the problematic image to the client, in another method, the server can also send the signature information of the problematic image to the client, namely the aforementioned MD5 information. By sending the MD5 information of the problematic image, the transmission speed can be improved, and it is convenient for the client to compare the MD5 information of the image to be reviewed with the MD5 information of the problematic image.

[0072] In one implementation where the server sends the MD5 hash of a problematic image to the client, the server can send a byte file containing MD5 hashes to the client. To improve data transmission security, the file sent by the server to the client can be encrypted. To improve data transmission speed, the file sent by the server to the client can be compressed. Each line in the file can correspond to the MD5 hash of a problematic image. To avoid an excessive number of MD5 hashes in the file, which could consume too much memory on the client side, an upper limit can be set on the number of MD5 hashes in the file, for example, a maximum of 10,000.

[0073] In one approach, to determine the issue level, cause of the issue, number of interactions, and other information for each image to be reviewed, the byte file sent by the server to the client may also contain information such as the issue level, cause of the issue, and number of interactions for each image.

[0074] Following the above S101, the image sending method provided in this embodiment further includes:

[0075] S102: Based on the signature information of the image to be reviewed, search for a target image that matches the image to be reviewed in the target image library.

[0076] In this embodiment of the disclosure, the target image library may store images sent by various clients. The images stored in the target image library may include images at various problem levels. Each problem level image may include images without problems or images with problems. In one approach, the target image library may store the MD5 hash of each image.

[0077] To facilitate the server's review of images to be reviewed, in one method, the server obtains the MD5 information corresponding to the image to be reviewed from the client.

[0078] Based on the MD5 hash of the image to be reviewed, a target image matching the MD5 hash of the image to be reviewed can be searched in the target image library. In other words, it can be checked whether a target image with a similarity threshold to the image to be reviewed exists. Alternatively, the hash value of the MD5 hash of the image to be reviewed can be compared with the hash values ​​of the MD5 hashes of various images in the target image library. In practice, it's possible that a target image matching the MD5 hash of the image to be reviewed will be searched in the target image library, or it's possible that no matching target image will be found in the target image library.

[0079] Following the above S102, the image sending method provided in this embodiment further includes:

[0080] S103: In response to finding a target image that matches the image to be reviewed, determine whether the image to be reviewed is a problematic image based on the problem level of the target image stored in the target image library.

[0081] Here, if a target image with a similarity to the image to be reviewed reaches a set threshold, it can be further determined whether the image to be reviewed is a problematic image.

[0082] For example, the problem levels of target images stored in the target image library can be arranged in ascending order of numbers. Here, 0 can be set to represent that the image has no problem; the larger the number, the higher the corresponding problem level.

[0083] S104: If it is determined that the image to be reviewed is not a problematic image, the image to be reviewed is sent to another client.

[0084] For example, in an instant messaging scenario, when the image to be reviewed is a group image, the server can send the image to all other clients in the target communication group to which the aforementioned client belongs. In a specific implementation, when the server receives the image to be reviewed from the client, it can determine the target communication group to which the aforementioned client belongs based on the obtained group identifier information, and then determine the other clients in the target communication group. When the image to be reviewed is a private image, when the server receives the image to be reviewed from the client, it can send the image to the target client corresponding to the aforementioned client based on the obtained identifier information of the target client used to receive the image.

[0085] Following on from S103 above, if it is determined that the image to be reviewed is a problematic image, the server cannot send the image to other clients. In one approach, the server can send a notification to the client indicating that the image to be reviewed is problematic, thus informing the user that the image cannot be sent and reducing interactions with problematic images.

[0086] Following the above S102, in one embodiment, the image sending method provided by this disclosure further includes:

[0087] If no matching target image is found, the image to be reviewed is sent to another client, and a pre-trained problem identification model is used to extract image features from the image to be reviewed. Based on the extracted image features, the problem level of the image to be reviewed is determined. Alternatively, if no matching target image is found, the text information contained in the image to be reviewed is extracted, and the extracted text information is compared with preset sensitive text to obtain the comparison result. Based on the comparison result, the problem level of the image to be reviewed is determined. The image to be reviewed and the determined problem level are then saved to the target image library.

[0088] No matching target image was found, meaning that the target image database may not contain any images whose similarity to the image under review meets the set threshold. To avoid disrupting the normal transmission of image information while preventing large-scale interactions when the image under review is problematic, asynchronous review can be implemented. This means that after sending the image to other clients, the server determines the issue level of the image under review.

[0089] Specifically, this can include the following three implementation methods: (i) In response to the failure to find a target image that matches the image to be reviewed, the image to be reviewed is sent to another client, and based on a pre-trained problem recognition model, the image features in the image to be reviewed are extracted, and based on the extracted image features, the problem level of the image to be reviewed is determined; the image to be reviewed and the determined problem level are saved to the target image library; (ii) In response to the failure to find a target image that matches the image to be reviewed, the image to be reviewed is sent to another client, and the text information contained in the image to be reviewed is extracted, the extracted text information is compared with preset sensitive text to obtain the comparison result, and based on the comparison result, the problem level of the image to be reviewed is determined; the image to be reviewed and the determined problem level are saved to the target image library. (iii) If no matching target image is found, the image to be reviewed is sent to another client. Using a pre-trained problem identification model, the image features in the image to be reviewed are extracted. Based on the extracted image features, the problem level of the image to be reviewed is determined. The text information contained in the target image is also extracted. The extracted text information is compared with preset sensitive text to obtain the comparison result. Based on the comparison result, the problem level of the image to be reviewed is determined. The image to be reviewed and the determined problem level are saved to the target image library.

[0090] The first implementation method described above can be an implementation method that, for images containing only non-textual information, determines the issue level and saves the image to be reviewed and the determined issue level to a target image library. Here, the image to be reviewed can be input into a pre-trained issue recognition model. The pre-trained issue recognition model can extract image features from the image to be reviewed, determine the issue level of the image to be reviewed based on the image features, and then output the issue level of the image to be reviewed. The pre-trained issue recognition model can be trained using samples containing image features and issue levels; the details of the issue recognition model will not be elaborated here.

[0091] The second implementation method described above can be an implementation method that, for an image to be reviewed containing text, determines the issue level and saves the image to be reviewed and the determined issue level to a target image library. Here, a text recognition algorithm can be used to extract the text information contained in the image to be reviewed, such as Optical Character Recognition (OCR). A preset correspondence between sensitive text and issue levels can be set in advance. Based on the comparison result between the extracted text information and the preset sensitive text, and the preset correspondence between sensitive text and issue levels, the issue level of the extracted text information can be determined, and thus the issue level of the image to be reviewed can be determined.

[0092] The third implementation method described above can be applied to images to be reviewed that contain both textual and non-textual information. For the non-textual information portion, the first implementation method described above can be used to determine the corresponding issue level; for the textual information portion, the second implementation method described above can be used to determine the corresponding issue level. Then, by combining the issue levels corresponding to the non-textual information and the issue levels corresponding to the textual information, the issue level of the image to be reviewed is determined.

[0093] In the asynchronous review process, in addition to determining the issue level of the image to be reviewed, the reasons for the issue and the level of interaction can also be determined. The reasons for the issue can be determined based on the image features extracted from the image and / or the text information contained within it. The level of interaction can be determined based on the obtained interaction details of the image, which include at least one of the following: number of interactions, number of participants in the interaction group, number of saves, and number of favorites. This process will be detailed below.

[0094] The number of interactions, the number of people in the interaction group, the number of saves, and the number of favorites are all positively correlated with the popularity of the interaction. That is, the more interactions, or the more people in the interaction group, or the more saves, or the more favorites, the higher the popularity of the interaction.

[0095] Information such as the issue level, reason for the issue, and interaction popularity of the images awaiting review can be stored in a relational database management system, such as MySQL, written in Structured Query Language (SQL). This information can also be stored in a key-value store database, such as a remote dictionary server (Redis). In Redis, the MD5 hash of the image to be reviewed can be used as the key, and the issue level, reason for the issue, and interaction popularity can be used as the value. Other methods include storing the creation and modification times of each image to be reviewed, which will not be detailed here. Using Redis can improve the server's parallel processing capability for images awaiting review.

[0096] The above process can be the initial review process of images to be reviewed by the server. To reduce problems with images during interaction, subsequent processes can also re-review each image stored in the target image library. Specifically, in one approach, firstly, the interaction details of each image in the target image library are obtained; the interaction details include at least one of the following: number of interactions, number of participants in the interaction group, number of saves, and number of favorites; then, based on the interaction details of each image, the interaction popularity of each image is determined; then, for each image, if the interaction popularity of the image reaches a set threshold, the problem level of the image is re-reviewed, and the problem level after re-review is determined; the re-review includes manual review and / or review using a model with more data than the initial review; finally, the problem level after re-review is saved to the target image library.

[0097] Here, interaction details can refer to the detailed information of each image during the interaction process. Specifically, it can include at least one of the following: number of interactions, number of people in the interaction group, number of saves, and number of favorites. Here, the number of people in the interaction group can refer to the number of people in the target communication group to which the client is located.

[0098] For example, when the interaction details include the number of interactions, the number of people in the interaction group, and the number of saves, the weight W of each image can be calculated using the following formula: W = T*10 + R*1 + S*5; where T represents the number of interactions; R represents the number of people in the interaction group; and S represents the number of saves. Then, the interaction popularity of each image is determined based on the weight W. The larger the weight W, the higher the interaction popularity of the image, and the higher the possibility of widespread interaction with the problematic image. Here, the weight W of each image can be continuously calculated. When the weight W of an image reaches a set threshold, a re-review of the image's problem level can be triggered. In other methods, the interaction popularity of each image can also be determined based on at least one of the following: the number of interactions, the number of people in the interaction group, the number of saves, and the number of favorites, thereby determining the problem level after re-review. Repeated methods will not be elaborated further.

[0099] In particular, when using more models than in the initial review, the review process is more rigorous and takes longer due to the larger number of models used. This increases the likelihood of detecting the cause of the problem and thus improves the accuracy of determining the problem level of each image.

[0100] Finally, the re-approved issue level can be saved to the target image library, thereby updating the issue level of each image in the target image library.

[0101] In other implementations, multiple different threshold values ​​can be set. As the interaction details of each image in the target image library are updated, each image can be re-evaluated when its interaction popularity reaches different threshold values. By re-evaluating each image multiple times, the accuracy of the issue level can be further improved.

[0102] As mentioned earlier, a problem image can refer to an image whose interaction count meets a first preset condition and whose problem level meets a second preset condition. Therefore, in one approach, it is possible to determine whether each image in the target image library is a problem image based on the interaction count and the problem level.

[0103] Specifically, firstly, the interaction count and issue level of each image in the target image library can be obtained; then, for each image, if the number of interactions of the image meets the first preset condition and the issue level of the image meets the second preset condition, the image is determined to be an issue image, and upon receiving an issue image update request from the client, the issue image is sent to the client, and the issue image is used by the client to update the stored issue images.

[0104] The more interactions an image receives and the higher its issue level, the more likely it is to be a problem image. Here, we can set a first preset condition as a threshold for the number of interactions and a second preset condition as a threshold for the issue level. For example, we can define images with more than 100 interactions and an issue level greater than or equal to 1 (where 0 represents no issue, and a higher value indicates a higher issue level) as problem images.

[0105] The issue level of an image can be the result of an initial review by the server, or the result of a re-review of all images in the target image library. In one approach, if an image has only undergone an initial review, the issue level from the initial review result, along with the number of interactions, is used to determine if it is an issue image. If the image has undergone a re-review, the issue level from the re-review result, along with the number of interactions, is used to determine if it is an issue image.

[0106] Here, upon receiving a request to update the problematic image from the client, the problematic image can be sent to the client. This process can refer to the process described in S101 above, and the repetitive parts will not be repeated.

[0107] In other implementations, the interaction popularity and issue level of each image in the target image library can also be obtained; then, for each image, if the interaction popularity of the image meets the third preset condition and the issue level of the image meets the second preset condition, the image is determined to be an issue image, and upon receiving an issue image update request sent by the client, the issue image is sent to the client, and the issue image is used by the client to update the stored issue images.

[0108] Similarly, the higher the interaction popularity of an image and the higher its issue level, the greater the likelihood that the image is a problematic image. Here, a third preset condition can be set as the image's interaction popularity threshold. The process of determining whether an image is a problematic image can refer to the above process, and repeated parts will not be elaborated upon.

[0109] The image sending method provided in this embodiment will be described below, taking the client as the executing entity as an example.

[0110] See Figure 2 The diagram shows a flowchart of another image sending method provided in this embodiment of the present disclosure. The method includes steps S201 to S203, wherein:

[0111] S201: Get the image to be sent.

[0112] Here, the image to be sent can be an image selected by the client user that has not been approved.

[0113] S202: Compare the image to be sent with the first problem image stored locally to determine whether the image to be sent matches the first problem image.

[0114] In this embodiment, the signature information, i.e., MD5 information, of the first problematic image can be stored locally. After obtaining the image to be sent, in one approach, the MD5 information of the image to be sent can be generated, and then the hash value of the MD5 information of the image to be sent can be compared with the hash value of the MD5 information of the first problematic image stored locally. The comparison result determines whether the first problematic image has not been matched. To reduce the size of the image data and save data storage space, in one approach, after obtaining the image to be sent, the image to be sent can be cropped to obtain a cropped image, and then the MD5 information corresponding to the cropped image can be generated. The image to be sent can be cropped to a preset image size according to specific cropping requirements; no specific limitation is made here.

[0115] In this embodiment of the disclosure, the first problem image stored locally may be sent from the server. Exemplarily, the first problem image may be sent proactively by the server; alternatively, it may be sent by the server in response to a problem image update request after the server sends such a request. Exemplarily, in one approach, the first problem image may be received from the server in response to a problem image update request and stored locally.

[0116] Specifically, a problem image update request is sent to the server; then, a second problem image is received from the server in response to the problem image update request; then, if the sum of the number of the second problem image and the first problem image exceeds a set threshold, the target problem image to be deleted is determined based on the duration information of the first problem image not being matched, and the target problem image is deleted.

[0117] To ensure the first problem image is updated promptly, an update request can be sent to the server in real time. To reduce data transmission pressure on both the server and client, in one approach, the client can send the update request to the server periodically, for example, once per hour. Each time the second problem image is received from the server, to avoid excessive memory consumption by the locally stored first problem image, the second problem image can be merged with the first problem image.

[0118] Here, you can set a threshold for the number of first-issue images. Upon receiving a second-issue image from the server, you can determine if the sum of the second and first-issue images exceeds the threshold. If it does, you need to delete the excess first-issue images to ensure that the sum of the second and remaining first-issue images does not exceed the threshold.

[0119] The duration for which the first problem image was not matched can refer to the time elapsed since the last time the first problem image was matched. In practice, the target problem images to be deleted can be determined based on the length of time they were not matched.

[0120] Considering that there might be instances where the duration of unmatched questions in the first set of question images is the same, in this case, the interaction popularity and / or question level of the first set of question images can be combined to determine the target question image to be deleted. In one approach, the first set of question images with the longest unmatched duration can be identified as a candidate question image to be deleted; it is then determined whether the sum of the remaining first set of question images and the second set of question images is less than a set threshold; if it is less, the target question image to be deleted from the candidate question images is determined based on the interaction popularity and / or question level of the candidate question images, until, after deleting the target question image, the sum of the remaining first set of question images and the second set of question images equals the set threshold.

[0121] For example, the local storage can hold a maximum of 10,000 MD5 hashes for the first problem images. When the total number of MD5 hashes for the first and second problem images is 10,005, and the longest-lasting first problem image that failed to match has 7 MD5 hashes, then to maintain the 10,000 MD5 hashes stored locally, 5 of these first problem images need to be deleted. If all 7 longest-lasting first problem images that failed to match are deleted, the total number of remaining first and second problem images will be less than 10,000.

[0122] Therefore, at this point, the seven longest-lasting first-question images that were not matched can be identified as candidate question images to be deleted. Then, based on the interaction popularity and / or question level of the seven candidate question images, five target question images to be deleted can be identified from the seven candidate question images.

[0123] Here, five target problem images to be deleted can be determined based on the interaction popularity of the seven candidate problem images; or five target problem images to be deleted can be determined based on the problem level of the seven candidate problem images; or five target problem images to be deleted can be determined based on both the interaction popularity and the problem level of the seven candidate problem images.

[0124] For example, target problem images to be deleted can be determined according to their interaction popularity in descending order and / or their problem level in descending order. For example, when determining target problem images to be deleted based on the interaction popularity and problem level of 7 candidate problem images, the weight information of each candidate problem image can be calculated. The weight information can be determined by the interaction popularity and problem level. Then, 5 target problem images to be deleted are determined according to their weights in descending order.

[0125] In one approach, if the sum of the remaining first and second problem images (excluding candidate problem images) exceeds a set threshold, the process can continue by identifying the longest-lasting first problem image among the remaining first problem images that was not matched as a candidate problem image to be deleted, until the sum of the remaining first and second problem images equals the set threshold. This process can be referenced from the previous process, and repetitions will not be repeated.

[0126] To avoid excessive memory consumption by the locally stored first problem image, one approach is to base the deletion on the similarity between the first and second problem images. If the similarity exceeds a set threshold, either the first or second problem image is identified as the target problem image for deletion, and then deleted. Here, the similarity can be determined based on the hash values ​​of the signature information of the first and second problem images; this process will not be detailed here.

[0127] S203: If the image to be sent does not match the first problematic image, the image to be sent to the server for further review.

[0128] If the image to be sent does not match the first problematic image, it indicates that the image was approved on the client's local machine, and can now be reviewed again on the server. The process of reviewing the image on the server can be referenced from steps S101 to S104; repeated parts will not be described further.

[0129] If the image to be sent matches the first problematic image, the image cannot be sent to the server. In this case, the image message can only be seen on the user's own client and cannot be sent to other clients.

[0130] Through the processes described in S201 to S203, the images to be sent can be reviewed locally on the client side, which can reduce the server's review pressure on problematic images to a certain extent.

[0131] Those skilled in the art will understand that, in the above-described method of the specific implementation, the order in which each step is written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined by its function and possible internal logic.

[0132] Based on the same inventive concept, this disclosure also provides embodiments similar to... Figure 1 The image sending method shown corresponds to an image sending device. Since the principle of the device in this embodiment solves the problem is the same as that in this embodiment... Figure 1The above-described image sending method is similar, therefore the implementation of the device can be found in the implementation of the method, and repeated details will not be repeated.

[0133] Reference Figure 3 The diagram shown is a schematic representation of the architecture of an image sending device according to an embodiment of this disclosure. The device includes: a receiving module 301, a searching module 302, a first determining module 303, and a sending module 304; wherein,

[0134] The receiving module 301 is used to receive an image to be reviewed sent by the client; the image to be reviewed is an image that the client determines does not match the problem image after comparing the image to be sent with the problem image stored in the client.

[0135] The search module 302 is used to search for a target image that matches the image to be reviewed from the target image library based on the signature information of the image to be reviewed;

[0136] The first determining module 303 is used to respond to finding a target image that matches the image to be reviewed, and determine whether the image to be reviewed is a problem image based on the problem level of the target image stored in the target image library;

[0137] The sending module 304 is used to send the image to be reviewed to other clients if it is determined that the image to be reviewed is not a problematic image.

[0138] In one feasible implementation, after the search module 302 performs the step of searching for a target image that matches the image to be reviewed from the target image library, the apparatus further includes:

[0139] The second determining module is configured to, in response to the absence of a matching target image, send the image to be reviewed to another client, extract image features from the image to be reviewed using a pre-trained problem identification model, and determine the problem level of the image to be reviewed based on the extracted image features; and / or, in response to the absence of a matching target image, send the image to be reviewed to another client, extract text information from the image to be reviewed, compare the extracted text information with preset sensitive text to obtain a comparison result, and determine the problem level of the image to be reviewed based on the comparison result;

[0140] The first saving module is used to save the image to be reviewed and the determined problem level to the target image library.

[0141] In one feasible implementation, the device further includes:

[0142] The first acquisition module is used to acquire interaction details information of each image in the target image library; the interaction details information includes at least one of the following: number of interactions, number of people in the interaction group, number of saves, and number of favorites;

[0143] The third determining module is used to determine the interaction popularity of each image based on the interaction details information of each image;

[0144] The fourth determination module is used to re-examine the problem level of each image when the interaction popularity of the image reaches a set threshold, and determine the problem level after re-examination; the re-examination includes manual review and / or review using more models than the initial review.

[0145] The second saving module is used to save the re-reviewed issue level to the target image library.

[0146] In one feasible implementation, the device further includes:

[0147] The second acquisition module is used to acquire the number of interactions and the issue level of each image in the target image library;

[0148] The fifth determining module is used to determine that each image is a problem image if the number of interactions with the image meets a first preset condition and the problem level of the image meets a second preset condition. Upon receiving a problem image update request from the client, the module sends the problem image to the client, and the problem image is used by the client to update the stored problem images.

[0149] Based on the same inventive concept, this disclosure also provides embodiments similar to... Figure 2 The image sending method shown corresponds to an image sending device. Since the principle of the device in this embodiment solves the problem is the same as that in this embodiment... Figure 2 The above-described image sending method is similar, therefore the implementation of the device can be found in the implementation of the method, and repeated details will not be repeated.

[0150] Reference Figure 4 The diagram shown illustrates the architecture of another image sending device provided in this embodiment of the present disclosure. The device includes: an acquisition module 401, a determination module 402, and a first sending module 403; wherein,

[0151] Module 401 is used to acquire the image to be sent;

[0152] The determining module 402 is used to compare the image to be sent with the first problem image stored locally to determine whether the image to be sent matches the first problem image;

[0153] The first sending module 403 is used to send the image to be sent to the server for re-review if the image to be sent does not match the first problem image.

[0154] In one feasible implementation, the device further includes:

[0155] The second sending module is used to send a problem image update request to the server;

[0156] The receiving module is used to receive the second problem image sent by the server in response to the problem image update request;

[0157] The deletion module is used to determine the target problem image to be deleted based on the duration information of when the sum of the number of the second problem image and the first problem image exceeds a set threshold, and then delete the target problem image.

[0158] In one feasible implementation, the deletion module is specifically used for:

[0159] The longest unmatched first question image among the first question images is identified as a candidate question image to be deleted;

[0160] Determine whether the sum of the number of remaining first question images (excluding the candidate question images) and the number of second question images is less than the set threshold;

[0161] If the threshold is less than the threshold, then based on the interaction popularity and / or issue level of the candidate issue images, a target issue image to be deleted is determined from the candidate issue images, until the sum of the number of remaining first issue images and second issue images equals the set threshold after deleting the target issue image.

[0162] In one feasible implementation, the device further includes:

[0163] The execution module is configured to, if the number of images is greater than the threshold, after deleting the candidate problem images, continue to execute the step of determining the longest unmatched first problem image among the remaining first problem images as the candidate problem image to be deleted, until the sum of the number of remaining first problem images and the number of second problem images equals the set threshold.

[0164] In one feasible implementation, the determining module 402 is specifically used for:

[0165] Generate signature information for the image to be sent, compare the hash value of the signature information of the image to be sent with the hash value of the signature information of the first problem image stored locally, and determine whether the image to be sent matches the first problem image.

[0166] The processing flow of each module in the device and the interaction flow between each module can be referred to the relevant descriptions in the above method embodiments, and will not be detailed here.

[0167] Based on the same technical concept, this disclosure also provides a computer device. (See also...) Figure 5 The diagram shows the structure of a computer device 500 provided in this embodiment of the present disclosure, including a processor 501, a memory 502, and a bus 503. The memory 502 stores execution instructions and includes main memory 5021 and external memory 5022. The main memory 5021, also called internal memory, is used to temporarily store computational data in the processor 501 and data exchanged with external memory 5022 such as a hard disk. The processor 501 exchanges data with the external memory 5022 through the main memory 5021. When the computer device 500 is running, the processor 501 and the memory 502 communicate through the bus 503, causing the processor 501 to execute the following instructions:

[0168] Receive images to be reviewed sent by the client; the images to be reviewed are those that the client determines do not match the problem images after comparing the images to be sent with the problem images stored in the client.

[0169] Based on the signature information of the image to be reviewed, a target image matching the image to be reviewed is searched from the target image library;

[0170] Upon finding a target image that matches the image to be reviewed, the system determines whether the image to be reviewed is a problematic image based on the problem level of the target image stored in the target image library.

[0171] If it is determined that the image to be reviewed is not a problematic image, the image to be reviewed will be sent to other clients.

[0172] Based on the same technical concept, this disclosure also provides another computer device. (Refer to...) Figure 6 The diagram shows the structure of a computer device 600 provided in this embodiment of the present disclosure, including a processor 601, a memory 602, and a bus 603. The memory 602 stores execution instructions and includes main memory 6021 and external memory 6022. The main memory 6021, also called internal memory, is used to temporarily store computational data in the processor 601 and data exchanged with external memory 6022 such as a hard disk. The processor 601 exchanges data with the external memory 6022 through the main memory 6021. When the computer device 600 is running, the processor 601 and the memory 602 communicate through the bus 603, causing the processor 601 to execute the following instructions:

[0173] Get the image to be sent;

[0174] The image to be sent is compared with the first problem image stored locally to determine whether the image to be sent matches the first problem image;

[0175] If the image to be sent does not match the first problematic image, the image to be sent to the server for further review.

[0176] This disclosure also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the image sending method described in the above method embodiments. The storage medium can be a volatile or non-volatile computer-readable storage medium.

[0177] This disclosure also provides a computer program product carrying program code. The program code includes instructions that can be used to execute the steps of the image sending method described in the above method embodiments. For details, please refer to the above method embodiments, which will not be repeated here.

[0178] The aforementioned computer program product can be implemented through hardware, software, or a combination thereof. In one optional embodiment, the computer program product is specifically embodied in a computer storage medium; in another optional embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0179] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the device described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here. In the several embodiments provided in this disclosure, it should be understood that the disclosed device and method can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection may be through some communication interfaces; the indirect coupling or communication connection of devices or units may be electrical, mechanical, or other forms.

[0180] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0181] In addition, the functional units in the various embodiments of this disclosure can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0182] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this disclosure. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0183] Finally, it should be noted that the above-described embodiments are merely specific implementations of this disclosure, used to illustrate the technical solutions of this disclosure, and not to limit it. The protection scope of this disclosure is not limited thereto. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this disclosure. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure, and should all be covered within the protection scope of this disclosure. Therefore, the protection scope of this disclosure should be determined by the protection scope of the claims.

Claims

1. A method for sending images, applied to a server, characterized in that, include: Receive images pending review sent by the client; The image to be reviewed is the image that the client determines does not match the problem image after comparing the image to be sent with the problem images stored in the client. Based on the signature information of the image to be reviewed, a target image matching the image to be reviewed is searched from the target image library; Upon finding a target image that matches the image to be reviewed, the system determines whether the image to be reviewed is a problematic image based on the problem level of the target image stored in the target image library. If it is determined that the image to be reviewed is not a problematic image, the image to be reviewed will be sent to other clients. The method further includes: If no matching image is found, the image to be reviewed is sent to the other client, and the server performs an initial asynchronous review of the image to determine its issue level and engagement level. If the interaction popularity reaches a set threshold, the issue level is re-evaluated, and the re-evaluated issue level is determined; the re-evaluated issue level is then saved to the target image library. The initial asynchronous review includes reviewing the image to be reviewed using a trained problem identification model, and the number of problem identification models used in the re-review is greater than the number of problem identification models used in the initial asynchronous review.

2. The method according to claim 1, characterized in that, After searching the target image library for a matching image to be reviewed, the method further includes: If no matching target image is found, the image to be reviewed is sent to another client, and a pre-trained problem identification model is used to extract image features from the image to be reviewed. Based on the extracted image features, the problem level of the image to be reviewed is determined; and / or, If no matching target image is found, the image to be reviewed is sent to another client, and the text information in the image to be reviewed is extracted. The extracted text information is compared with preset sensitive text to obtain a comparison result, and the problem level of the image to be reviewed is determined based on the comparison result. The images to be reviewed and the determined issue levels are saved to the target image library.

3. The method according to claim 1, characterized in that, The method further includes: Obtain interaction details information for each image in the target image library; the interaction details information includes at least one of the following: number of interactions, number of participants in the interaction group, number of saves, and number of favorites; Based on the interaction details of each image, the interaction popularity of each image is determined; The re-review includes manual review.

4. The method according to claim 1, characterized in that, The method further includes: Obtain the interaction count and issue level of each image in the target image library; For each image, if the number of interactions with the image meets a first preset condition and the problem level of the image meets a second preset condition, the image is determined to be a problem image. Upon receiving a problem image update request from the client, the problem image is sent to the client, and the problem image is used by the client to update the stored problem images.

5. A method for sending images, applied to a client, characterized in that, include: Get the image to be sent; The image to be sent is compared with the first problem image stored locally to determine whether the image to be sent matches the first problem image; If the image to be sent does not match the first problematic image, the image to be sent to the server for further review. The server, based on the signature information of the image to be sent, searches for a matching target image in the target image library. If no matching target image is found, the server sends the image to another client. The server also performs an initial asynchronous review of the image to be sent to determine its issue level and interaction popularity. This initial asynchronous review includes using a trained issue recognition model to review the image. Specifically, when the interaction popularity reaches a set threshold, the server re-examines the issue level and determines the re-examined issue level; the re-examined issue level is saved to the target image library, and the number of issue recognition models used in the re-examination is greater than the number of issue recognition models used in the initial asynchronous examination.

6. The method according to claim 5, characterized in that, The method further includes: Send a request to the server to update the problematic image; Receive the second problem image sent by the server in response to the problem image update request; If the sum of the number of the second problem image and the first problem image exceeds a set threshold, a target problem image to be deleted is determined based on the duration information of when the first problem image was not matched, and the target problem image is deleted.

7. The method according to claim 6, characterized in that, The step of determining the target problem image to be deleted based on the duration information of the first problem image not being matched includes: The longest unmatched first question image among the first question images is identified as a candidate question image to be deleted; Determine whether the sum of the number of remaining first question images (excluding the candidate question images) and the number of second question images is less than the set threshold; If the threshold is less than the threshold, then based on the interaction popularity and / or issue level of the candidate issue images, a target issue image to be deleted is determined from the candidate issue images, until the sum of the number of remaining first issue images and second issue images equals the set threshold after deleting the target issue image.

8. The method according to claim 7, characterized in that, After determining whether the sum of the number of remaining first question images (excluding the candidate question images) and the number of second question images is less than the set threshold, the method further includes: If the value is greater than the threshold, after deleting the candidate problem image, the step of determining the longest unmatched first problem image among the remaining first problem images as the candidate problem image to be deleted continues until the sum of the number of remaining first problem images and the number of second problem images equals the set threshold.

9. The method according to claim 5, characterized in that, The step of comparing the image to be sent with the first problem image stored locally to determine whether the image to be sent matches the first problem image includes: Generate signature information for the image to be sent, compare the hash value of the signature information of the image to be sent with the hash value of the signature information of the first problem image stored locally, and determine whether the image to be sent matches the first problem image.

10. A picture sending device, characterized in that, include: The receiving module is used to receive images to be reviewed sent by the client. The image to be reviewed is the image that the client determines does not match the problem image after comparing the image to be sent with the problem images stored in the client. The search module is used to search for a target image that matches the image to be reviewed from the target image library based on the signature information of the image to be reviewed. The first determining module is used to determine whether the image to be reviewed is a problematic image based on the problem level of the target image stored in the target image library when a target image matching the image to be reviewed is found. The sending module is used to send the image to be reviewed to other clients if it is determined that the image to be reviewed is not a problematic image. The image sending device is further configured to: in response to the absence of a matching target image, send the image to be reviewed to the other client, and have the server perform an initial asynchronous review of the image to determine its issue level and engagement level. The device further includes a fourth determining module, wherein the fourth determining module is configured to: re-examine the issue level when the interaction popularity reaches a set threshold, and determine the re-examined issue level; save the re-examined issue level to the target image library. The initial asynchronous review includes reviewing the image to be reviewed using a trained problem identification model, and the number of problem identification models used in the re-review is greater than the number of problem identification models used in the initial asynchronous review.

11. A picture sending device, characterized in that, include: The acquisition module is used to acquire images to be sent. The determination module is used to compare the image to be sent with the first problem image stored locally to determine whether the image to be sent matches the first problem image; The first sending module is configured to send the image to be sent to the server for further review if the image to be sent does not match the first problematic image. The server, based on the signature information of the image to be sent, searches for a matching target image in the target image library. If no matching target image is found, the server sends the image to another client. The server also performs an initial asynchronous review of the image to be sent to determine its issue level and interaction popularity. This initial asynchronous review includes using a trained issue recognition model to review the image. Specifically, when the interaction popularity reaches a set threshold, the server re-examines the issue level and determines the re-examined issue level; the re-examined issue level is saved to the target image library, and the number of issue recognition models used in the re-examination is greater than the number of issue recognition models used in the initial asynchronous examination.

12. A computer device, characterized in that, include: The computer device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, they perform the steps of the image sending method as described in any one of claims 1 to 4 or the steps of the image sending method as described in any one of claims 5 to 9.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the image sending method as described in any one of claims 1 to 4 or the steps of the image sending method as described in any one of claims 5 to 9.

Images