Image annotation method, device, electronic device and storage medium

Abstract

The embodiments of the present application provide an image annotation method, device, electronic device and storage medium, which relate to the field of image processing technology. The method comprises: first determining the workgroups to which multiple target accounts logged in in parallel belong respectively, then determining the annotation object corresponding to the workgroup to which the target account belongs based on the first correspondence between the workgroup and the annotation object, and generating an annotation task for the target account, then determining the designated annotation object for the target account, and based on the designated annotation object, retrieving the image to be annotated of the target account, thereby determining the image to be pushed of each target account, and then pushing the image to be pushed of each target account to the target terminal device, so that the target terminal device displays the image to be pushed, and finally obtaining the annotation information of the image to be pushed of the target account. By applying the embodiments of the present application, the accuracy of image annotation is improved while ensuring the efficiency of image annotation.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to an image annotation method, device, electronic device, and storage medium. Background Art

[0002] With the rapid development of computer technology, image processing technology has been gradually applied to various fields, and the demand for image processing models has also increased accordingly. Typically, sample images for training image processing models can be obtained by annotating image data. For example, objects in an image can be annotated by category to obtain sample images for training image classification models, and vehicles in an image can be annotated by category to obtain sample images for a vehicle classification model.

[0003] In related technologies, in order to improve the efficiency of image annotation, the annotation of various image data is usually completed by the same staff member. However, since the annotation of different image data may adopt different annotation methods and require different annotation content, annotation errors are prone to occur, resulting in low accuracy of the obtained sample images, and further, low accuracy of the trained image processing model.

[0004] Based on this, how to improve the accuracy of image annotation while ensuring the efficiency of image annotation is a technical problem that needs to be solved urgently. Summary of the Invention

[0005] The purpose of the embodiments of the present application is to provide an image annotation method, apparatus, electronic device, and storage medium to improve the accuracy of image annotation while ensuring image annotation efficiency. The specific technical solutions are as follows:

[0006] In a first aspect, an embodiment of the present application provides an image annotation method, the method comprising:

[0007] Determine the workgroups to which multiple target accounts logged in in parallel belong;

[0008] For each target account, determining the annotation object corresponding to the workgroup to which the target account belongs based on the first correspondence between the workgroup and the annotation object, and generating a annotation task for the target account based on the determined annotation object; wherein the annotation task for each target account includes: the annotation object corresponding to the workgroup to which the target account belongs;

[0009] For each target account, a designated annotation object for the target account is determined, and based on the designated annotation object, images to be annotated for the target account are retrieved; wherein the designated annotation object is the object indicated by the received object selection instruction in the annotation task for the target account, and the image to be annotated is an image of a terminal device in which the designated annotation object contained in a preset image is not annotated and has not been pushed outside a target terminal device, and the object indicated by the received object selection instruction is the designated annotation object; the target terminal device is the terminal device on which the target account is logged in;

[0010] Determine the images to be pushed to each target account based on the images to be annotated for each target account;

[0011] Pushing the image to be pushed of each target account to the target terminal device logged in with the target account, so that the target terminal device displays the received image to be pushed;

[0012] For each target account, obtain the annotation information of the image to be pushed of the target account; wherein the annotation information is sent by the target terminal device and is the annotation information of the specified annotation object in the image to be pushed of the target account.

[0013] Optionally, in a specific implementation method, the image to be pushed for each target account is determined based on the image to be annotated of each target account, including: determining the image to be annotated of each target account as the image to be pushed of the target account; or, determining the sub-image of the designated annotation object of the target account in the image to be annotated of each target account as the image to be pushed of the target account.

[0014] Optionally, in a specific implementation, the pushing of the images to be pushed of each target account to the target terminal device includes: for each target account, pushing a specified number of remaining images to the target terminal device; wherein, the remaining images are: the images to be pushed of the target account that have not been pushed to the target terminal device; the method also includes: for each target account, after obtaining the annotation information of all the images to be pushed that have been pushed of the target account, returning to the step of pushing the specified number of remaining images to the target terminal device until the target account terminates login, or the remaining images do not exist; or, for each target account, after obtaining the annotation information of the preset number of images to be pushed that have been pushed of the target account, pushing the preset number of remaining images to the target terminal device until the target account terminates login, or the remaining images do not exist.

[0015] Optionally, in a specific implementation method, for each target account, the annotation object corresponding to the workgroup to which the target account belongs is determined based on the first correspondence between the workgroup and the annotation object, including: for each target account, searching the workgroup to which the target account belongs in a pre-acquired annotation scheme, and searching the annotation object corresponding to the workgroup to which the target account belongs in the annotation scheme; wherein the annotation scheme includes: the first correspondence between the workgroup and the annotation object, and the subordinate relationship between the account and the workgroup.

[0016] Optionally, in a specific implementation, the annotation scheme also includes: a second correspondence between the annotation object, candidate annotation content and annotation method; before pushing the image to be annotated corresponding to each target account to the target terminal device, the method also includes: determining the candidate annotation content and annotation method of the designated annotation object of each target account according to the second correspondence in the annotation scheme, as the candidate annotation content and annotation method of the target account; pushing the image to be pushed of each target account to the target terminal device so that the target terminal device displays the received image to be pushed, including: pushing the image to be pushed of each target account carrying the candidate annotation content and annotation method of the target account to the target terminal device, so that the target terminal device displays the received image to be pushed, candidate annotation content and annotation method; obtaining the annotation information of the image to be pushed of the target account for each target account, including: obtaining the annotation information of the image to be pushed of the target account selected from the candidate annotation content of the target account according to the annotation method of the target account.

[0017] Optionally, in a specific implementation, the method further includes: in the first correspondence, adding a new entry regarding the correspondence between the workgroup and the marked object, and / or deleting the current entry regarding the object relationship between the workgroup and the marked object, and returning to the step of determining multiple target accounts for parallel login.

[0018] Optionally, in a specific implementation, the method further includes: adding a new entry regarding the subordinate relationship between the account and the workgroup in the marking scheme, and / or deleting the current entry regarding the subordinate relationship between the account and the workgroup, and returning to the step of determining multiple target accounts for parallel login.

[0019] Optionally, in a specific implementation, the method further includes: determining the target annotation information of each target image that has completed information annotation in the preset image, and integrating the target annotation information of each target image to obtain each target image carrying the target integration information; and sending each target image carrying the target integration information to a designated device.

[0020] In a second aspect, an embodiment of the present application provides an image annotation device, the device comprising:

[0021] A workgroup determination module is used to determine the workgroups to which multiple target accounts logged in in parallel belong;

[0022] The task generation module is configured to determine, for each target account, the annotation object corresponding to the workgroup to which the target account belongs based on the first correspondence between the workgroup and the annotation object, and generate an annotation task for the target account based on the determined annotation object, wherein the annotation task for each target account includes: the annotation object corresponding to the workgroup to which the target account belongs;

[0023] An image retrieval module is configured to determine, for each target account, a designated annotation object for the target account, and retrieve images to be annotated for the target account based on the designated annotation object; wherein the designated annotation object is the object indicated by the received object selection instruction in the annotation task for the target account; the image to be annotated is an image of a terminal device in which the designated annotation object contained in a preset image is not annotated and has not been pushed outside a target terminal device, and the object indicated by the received object selection instruction is the designated annotation object; the target terminal device is the terminal device on which the target account is logged in;

[0024] An image determination module, configured to determine the image to be pushed to each target account based on the image to be annotated of each target account;

[0025] An image pushing module, configured to push the image to be pushed of each target account to the target terminal device, so that the target terminal device displays the received image to be pushed;

[0026] The information acquisition module is used to obtain the annotation information of the image to be pushed of each target account; wherein the annotation information is sent by the target terminal device and is the annotation information of the specified annotation object in the image to be pushed of the target account.

[0027] Optionally, in a specific implementation method, the image determination module is specifically used to: determine the image to be annotated of each target account as the image to be pushed of the target account; or, determine the sub-image of the target annotation object corresponding to the target account in the image to be annotated of each target account as the image to be pushed of the target account.

[0028] Optionally, in a specific implementation, the image push module is specifically configured to: for each target account, push a specified number of remaining images to the target terminal device; wherein the remaining images are: images of the target account that have not been pushed to the target terminal device and are to be pushed; the apparatus further comprises:

[0029] The remaining image pushing module is used to trigger the image pushing module for each target account after obtaining the annotation information of all the images to be pushed that have been pushed by the target account, until the target account terminates login, or there are no remaining images; or, for each target account, after obtaining the annotation information of a preset number of images to be pushed that have been pushed by the target account, push the preset number of remaining images to the target terminal device until the target account terminates login, or there are no remaining images.

[0030] Optionally, in a specific implementation method, the task generation module is specifically used to: for each target account, retrieve the workgroup to which the target account belongs in a pre-acquired labeling scheme, and retrieve the labeling object corresponding to the workgroup to which the target account belongs in the labeling scheme; wherein the labeling scheme includes: a first correspondence between the workgroup and the labeling object, and a subordinate relationship between the account and the workgroup.

[0031] Optionally, in a specific implementation, the annotation scheme further includes: a second correspondence between the annotation object, the candidate annotation content, and the annotation method; and the device further includes:

[0032] a content and method determination module configured to determine, before pushing the to-be-annotated image corresponding to each target account to the target terminal device, candidate annotation content and annotation method for the designated annotation object of each target account based on the second corresponding relationship in the annotation scheme, as the candidate annotation content and annotation method for the target account;

[0033] The image push module is specifically configured to push the image to be pushed, which carries the candidate annotation content and annotation method of each target account, to the target terminal device, so that the target terminal device displays the received image to be pushed, the candidate annotation content, and the annotation method;

[0034] The information acquisition module is specifically configured to: for each target account, acquire, according to the target account's annotation method, the annotation information of the target account's to-be-pushed image selected from the target account's candidate annotation content.

[0035] Optionally, in a specific implementation, the device further includes:

[0036] The first module for adding and deleting entries is used to add new entries about the correspondence between the workgroup and the marked object in the first correspondence, and / or delete the current entries about the object relationship between the workgroup and the marked object, and return to the step of determining multiple target accounts for parallel login.

[0037] Optionally, in a specific implementation, the device further includes:

[0038] The second module for adding and deleting entries is used to add new entries about the subordinate relationship between the account and the workgroup in the marking scheme, and / or delete the current entry about the subordinate relationship between the account and the workgroup, and return to the step of determining multiple target accounts for parallel login.

[0039] Optionally, in a specific implementation, the device further includes:

[0040] An information integration module is used to determine target annotation information of each target image that has completed information annotation in the preset image, and integrate the target annotation information of each target image to obtain each target image carrying target integration information;

[0041] The image sending module is used to send each target image carrying target integration information to a designated device.

[0042] In a third aspect, an embodiment of the present application provides an electronic device, including:

[0043] Memory for storing computer programs;

[0044] The processor is configured to implement the steps of any one of the image annotation methods provided in the first aspect when executing the program stored in the memory.

[0045] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of any image annotation method provided in the first aspect are implemented.

[0046] In a fifth aspect, an embodiment of the present application further provides a computer program product comprising instructions, which, when executed on a computer, enables the computer to execute any of the image annotation methods described in the first aspect above.

[0047] Beneficial effects of the embodiments of the present application:

[0048] As can be seen from the above, by applying the technical solution provided in the embodiment of the present application, the workgroups to which the multiple target accounts logged in in parallel belong are first determined, and then for each target account, the annotation object corresponding to the workgroup to which the target account belongs is determined based on the first correspondence between the workgroup and the annotation object, and based on the determined annotation object, the annotation task of the target account is generated, and then for each target account, the designated annotation object of the target account is determined, and based on the designated annotation object, the image to be annotated of the target account is retrieved, and then based on the image to be annotated of each target account, the image to be pushed of each target account is determined, and then the image to be pushed of each target account is pushed to the target terminal device logged in with the target account, so that the terminal device displays the received image to be pushed, and finally, for each target account, the annotation information of the image to be pushed of the target account is obtained.

[0049] Based on this, for multiple objects that need to be labeled, the objects to be labeled are assigned to different workgroups, thereby establishing a first correspondence between the workgroup and the labeled object. Multiple accounts for image annotation are assigned to the workgroup. After the staff member logs in to the account, the workgroup to which the account belongs is determined, and the labeled object is determined according to the first correspondence, thereby completing the labeling of the labeled object. In this way, when labeling, different staff members can log in to the account in parallel, so that the corresponding images to be labeled can be labeled in parallel. Since multiple accounts are logged in in parallel, multiple labeled objects of multiple images can be labeled at the same time, so the efficiency of image labeling can be guaranteed. Moreover, since different staff members' accounts are assigned different labeled objects, the staff member assigned to each labeled object does not need to pay attention to all the labeled objects. Therefore, it is not easy for the staff member to make mistakes when labeling the image. While ensuring the efficiency of image labeling, the accuracy of image labeling can be improved, thereby improving the accuracy of image samples, and further improving the accuracy of subsequent image model training. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other embodiments can also be obtained based on these drawings.

[0051] Figure 1 A schematic flow chart of the first image annotation method provided in an embodiment of the present application.

[0052] Figure 2a A schematic flow chart of the second image annotation method provided in an embodiment of the present application.

[0053] Figure 2b A schematic diagram of the process of the third image annotation method provided in an embodiment of the present application.

[0054] Figure 3 A schematic flow chart of the fourth image annotation method provided in an embodiment of the present application.

[0055] Figure 4 A flowchart of the fifth image annotation method provided in an embodiment of the present application.

[0056] Figure 5 A schematic diagram of the structure of an image annotation device provided in an embodiment of the present application.

[0057] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field based on this application are within the scope of protection of this application.

[0059] In related technologies, to improve the efficiency of image annotation, the same staff member is usually responsible for annotating multiple image data. However, because different image data may be annotated using different methods and require different annotation content, annotation errors are prone to occur, resulting in low accuracy of the sample images and, in turn, low accuracy of the trained image processing models. Therefore, how to improve the accuracy of image annotation while maintaining efficiency is a technical problem that needs to be solved urgently.

[0060] In order to solve the above technical problems, an embodiment of the present application provides an image annotation method.

[0061] This method can be applied to various application scenarios requiring image annotation, for example, obtaining sample images for image model training by annotating images. Furthermore, this method can be applied to various types of electronic devices, such as servers, desktop computers, and mobile phones (hereinafter referred to as electronic devices). Therefore, the embodiments of the present invention do not limit the application scenarios or execution entities of this method.

[0062] The electronic device can execute the image annotation method via an installed client with image annotation capabilities, or it can execute the image annotation method via a module configured with the image annotation function. Furthermore, when the electronic device executes the image annotation method via a client, the module for implementing the image annotation method can be a plug-in to the client; the client can also be a client specifically designed to execute the object detection method. All of these are reasonable.

[0063] An image annotation method provided in an embodiment of the present application may include the following steps:

[0064] Determine the workgroups to which multiple target accounts logged in in parallel belong;

[0065] For each target account, determining the annotation object corresponding to the workgroup to which the target account belongs based on the first correspondence between the workgroup and the annotation object, and generating a annotation task for the target account based on the determined annotation object; wherein the annotation task for each target account includes: the annotation object corresponding to the workgroup to which the target account belongs;

[0066] For each target account, a designated annotation object for the target account is determined, and based on the designated annotation object, images to be annotated for the target account are retrieved; wherein the designated annotation object is the object indicated by the received object selection instruction in the annotation task for the target account, and the image to be annotated is an image of a terminal device in which the designated annotation object contained in a preset image is not annotated and has not been pushed outside a target terminal device, and the object indicated by the received object selection instruction is the designated annotation object; the target terminal device is the terminal device on which the target account is logged in;

[0067] Determine the images to be pushed to each target account based on the images to be annotated for each target account;

[0068] Pushing the image to be pushed of each target account to the target terminal device, so that the target terminal device displays the received image to be pushed;

[0069] For each target account, obtain the annotation information of the image to be pushed of the target account; wherein the annotation information is sent by the target terminal device and is the annotation information of the specified annotation object in the image to be pushed of the target account.

[0070] As can be seen from the above, by applying the technical solution provided in the embodiment of the present application, the workgroups to which the multiple target accounts logged in in parallel belong are first determined, and then for each target account, the annotation object corresponding to the workgroup to which the target account belongs is determined based on the first correspondence between the workgroup and the annotation object, and based on the determined annotation object, the annotation task of the target account is generated, and then for each target account, the designated annotation object of the target account is determined, and based on the designated annotation object, the image to be annotated of the target account is retrieved, and then based on the image to be annotated of each target account, the image to be pushed of each target account is determined, and then the image to be pushed of each target account is pushed to the target terminal device logged in with the target account, so that the terminal device displays the received image to be pushed, and finally, for each target account, the annotation information of the image to be pushed of the target account is obtained.

[0071] Based on this, for multiple objects that need to be labeled, the objects to be labeled are assigned to different workgroups, thereby establishing a first correspondence between the workgroup and the labeled object. Multiple accounts for image annotation are assigned to the workgroup. After the staff member logs in to the account, the workgroup to which the account belongs is determined, and the labeled object is determined according to the first correspondence, thereby completing the labeling of the labeled object. In this way, when labeling, different staff members can log in to the account in parallel, so that the corresponding images to be labeled can be labeled in parallel. Since multiple accounts are logged in in parallel, multiple labeled objects of multiple images can be labeled at the same time, so the efficiency of image labeling can be guaranteed. Moreover, since different staff members' accounts are assigned different labeled objects, the staff member assigned to each labeled object does not need to pay attention to all the labeled objects. Therefore, it is not easy for the staff member to make mistakes when labeling the image. While ensuring the efficiency of image labeling, the accuracy of image labeling can be improved, thereby improving the accuracy of image samples, and further improving the accuracy of subsequent image model training.

[0072] Before specifically describing an image annotation method provided by an embodiment of the present invention, the first corresponding relationship involved in the method is described:

[0073] Typically, multiple objects in an image can be annotated. Thus, an image to be annotated can contain multiple annotated objects, such as faces, objects, vehicles, etc. Therefore, we can first determine the multiple annotated objects in the image to be annotated based on the annotation requirements, and assign staff members to annotate these objects.

[0074] Determine the annotation object and assign a workgroup for image annotation to each annotation object. Since image annotation is completed by the staff by logging into the account assigned to the workgroup on the terminal device to complete the annotation of the image to be annotated, the staff logs in to the corresponding account in the workgroup, and after determining the workgroup to which the account belongs, annotates the determined annotation object according to the first correspondence between the workgroup and the annotation object; wherein, the account logged in by the staff can be a fixed account or a randomly determined account, and this application does not make specific restrictions on this.

[0075] Due to the large number of images to be annotated, in order to improve the efficiency of image annotation, it is necessary to divide the annotated objects into multiple workgroups. The annotation objects corresponding to each workgroup are the same object. Each staff member can annotate the images corresponding to their own annotation objects by logging into an account in the workgroup. Therefore, each annotation object should correspond to at least one workgroup, so that a first correspondence between the workgroup and the annotation object can be established. The first correspondence represents the annotation permissions in the workgroup. For an account belonging to a workgroup included in the first correspondence, when a user logs in to the account, the user only has the annotation permission for the object corresponding to the account, and cannot annotate other objects that are not corresponding.

[0076] For example, there are two annotation objects: eyes and mouth. The annotation objects of workgroups A and B are eyes and mouth respectively. The account assigned to workgroup A is A, and the account assigned to workgroup B is B. Therefore, the eyes have a first correspondence with account A, and the mouth has a first correspondence with account B. Therefore, staff member C logs in to account A on the electronic device, so that he can annotate the eyes, and staff member D logs in to account B on the electronic device, so that he can annotate the mouth.

[0077] Optionally, a labeling scheme can be created that includes a first correspondence between a workgroup and an annotation object, wherein the correspondence between the first relationship between the workgroup and the annotation object in the labeling scheme can be by establishing an entry table with a one-to-one correspondence between the workgroup and the annotation object; or it can be by building a database of the correspondence between the workgroup and the annotation object.

[0078] Optionally, because different annotation objects may correspond to different candidate annotation contents and annotation methods, the annotation scheme may further include: a second correspondence between the annotation object, the candidate annotation contents, and the annotation method; wherein the second correspondence is determined as follows:

[0079] According to the annotation object, suitable candidate annotation content is selected for each annotation object, and then the most suitable annotation method is determined according to the difference between the candidate annotation content, thereby determining a second correspondence between the annotation object, the candidate annotation content and the annotation method;

[0080] The candidate annotation contents include: one or more of the shape, color, quantity and size of the annotation object; the annotation method includes: one or more of check mark and text mark;

[0081] Generally speaking, the presentation form of the second correspondence between the annotation objects, candidate annotation contents and annotation methods corresponding to the account can be expressed as a one-to-one correspondence table between the above-mentioned annotation objects, candidate annotation contents and annotation methods established in the annotation scheme, or it can be a database built in the annotation scheme containing the above-mentioned annotation objects, candidate annotation contents and annotation methods. The candidate annotation contents and annotation methods can be searched according to the corresponding annotation objects of the determined account, which is reasonable.

[0082] For example, the existing annotation objects are the category of vehicles and the color of vehicles, the candidate annotation contents are category 1, category 2, category 3 of vehicles and color a, color b, color c of vehicles, and the annotation methods are check mark and text mark; the annotation objects of staff members C and D are the category of vehicles and the color of vehicles respectively, the candidate annotation contents of C are category 1, category 2, category 3 of vehicles, and the candidate annotation contents of D are color a, color b, color c of vehicles. For the above candidate annotation contents, the most suitable annotation method is check mark, so the category of vehicles has a second corresponding relationship with category 1, category 2, category 3 and check mark, and the color of vehicles has a second corresponding relationship with color a, color b, color c and check mark. Therefore, staff member C logs in to account A on the electronic device and can use check mark to mark the category of vehicles, and staff member D logs in to account B on the electronic device and can use check mark to mark the color of vehicles.

[0083] Below, based on the above description and in combination with the accompanying drawings, an image annotation method provided in an embodiment of the present application is specifically described.

[0084] Figure 1 This is a flow chart of an image annotation method provided in an embodiment of the present application, such as Figure 1 As shown, the method includes steps S101-S106:

[0085] Step S101: Determine the workgroups to which multiple target accounts logged in in parallel belong respectively;

[0086] After the target accounts are assigned to various workgroups, multiple staff members can log in to their assigned target accounts in parallel, and the electronic device can detect that multiple accounts are logged in in parallel; thus, after the electronic device determines the multiple target accounts logged in in parallel, it can determine the workgroup to which each target account belongs.

[0087] When allocating target accounts, all target accounts can be evenly distributed to each workgroup; the number of target accounts contained in each workgroup can also be allocated according to actual needs. For example, when the annotation objects corresponding to certain workgroups are more difficult to label, the number of target accounts in the workgroup can be increased; for example, when the labeling completion time limit in some workgroups is relatively short, the number of target accounts in the workgroup can be increased. This is all reasonable.

[0088] In addition, the same target account may be assigned to one workgroup or multiple workgroups, and this application does not impose any specific restrictions on this.

[0089] Step S102: For each target account, determine the annotation object corresponding to the workgroup to which the target account belongs based on the first correspondence between the workgroup and the annotation object, and generate a annotation task for the target account based on the determined annotation object;

[0090] The labeling task for each target account includes: the labeling object corresponding to the workgroup to which the target account belongs;

[0091] The first correspondence between the workgroup and the annotation object is a one-to-one correspondence, that is, each workgroup corresponds to only one annotation object.

[0092] After the target account has established the annotation objects based on the first correspondence, a annotation task can be generated. The annotation task includes all the annotation objects that need to be annotated by the target account, so that the staff using the target account can clearly and intuitively understand their work tasks and select the annotation objects that need to be annotated first.

[0093] Optionally, the labeling task may include one or more of labeling objects, time limit requirements for each labeling object, and unlabeled images of each labeling object.

[0094] Step S103: for each target account, determining a designated annotation object for the target account, and retrieving the to-be-annotated images of the target account based on the designated annotation object;

[0095] Among them, the designated annotation object is: the object indicated by the received object selection instruction in the annotation task of the target account, the image to be annotated is: the designated annotation object contained in the preset image is not annotated, and has not been pushed outside the target terminal device and the object indicated by the received object selection instruction is the image of the terminal device of the designated annotation object, and the target terminal device is: the terminal device with which the target account is logged in.

[0096] For example, if the workgroups to which target accounts A and B belong both have eyes as the labeling object, and worker 1 logs in to target account A on terminal device 1 and receives an object selection instruction that specifies eyes as the labeling object, and worker 2 logs in to target account B on terminal device 2 and receives an object selection instruction that specifies eyes as the labeling object, then the same image to be labeled with eyes as the labeling object can be randomly pushed to any terminal device logged in to target account A or target account B. However, once an image to be labeled is pushed to an electronic device logged in to target account A, it cannot be pushed to an electronic device logged in to target account B, and vice versa.

[0097] For another example, if the workgroups to which target accounts A and B belong both have eyes and hair as the annotation objects, and worker 1 logs in to target account A on terminal device 1 and, according to the object selection instruction received, specifies eyes as the designated annotation object, and worker 2 logs in to target account B on terminal device 2 and, according to the object selection instruction received, specifies hair as the designated annotation object, then the same image to be annotated can be pushed to the terminal device logged in to target account A for eye annotation, or pushed to the terminal device logged in to target account B for hair annotation. In this case, the image to be annotated pushed to target account A can be omitted from the hair, and the image to be annotated pushed to target account B can be omitted from the eyes.

[0098] Optionally, when each target account belongs to only one workgroup, and the workgroup has only the same annotated object, that is, there is a one-to-one correspondence between the target account, the workgroup and the annotated object, the same annotated image containing the annotated object will be pushed to the terminal device logged in by any target account included in the workgroup, and will no longer be pushed to the terminal device logged in by other target accounts in the workgroup.

[0099] After receiving the above-mentioned annotation tasks, the staff using the target account can intuitively understand all the annotation objects they need to annotate; when there is only one annotation object in the annotation task list, the annotation object will be annotated by default; when there are multiple annotation objects in the annotation task, the staff can choose an annotation object to be annotated first. After the selection of the annotation object is determined, only the images to be annotated of the annotation object will be retrieved.

[0100] Optionally, after the staff determines the annotation object, the terminal device logged in by the target account used will only display the images to be annotated retrieved for the annotation object; the relevant information interface of the remaining annotation objects will be moved to a small window. After completing the annotation of the images to be annotated for the annotation object, a new annotation object can be determined to retrieve the images to be annotated.

[0101] Usually, when annotating images, multiple images that need to be annotated can be pre-set. Therefore, after determining the target annotation object corresponding to each target account, the image that needs to be annotated by the staff corresponding to the target account can be selected from the preset images to complete the image annotation.

[0102] Among them, the electronic device can retrieve the image to be annotated of each target account in the preset image, so as to facilitate the subsequent push of the image to be pushed to each target account; in order to ensure that the staff corresponding to the target account can annotate the target annotation object that needs to be annotated based on the first corresponding relationship, the target annotation object determined by the target account according to the first corresponding relationship is retrieved in the preset image to be in an unannotated state. At the same time, in order to avoid the same target annotation object being annotated repeatedly, when searching the preset image, it is also necessary to ensure that the target annotation object has not been pushed to other accounts that can also annotate the target annotation object.

[0103] Therefore, for each account, the image to be annotated retrieved by the electronic device is an image in which the target annotation object corresponding to the target account is not annotated and has not been pushed to terminal devices other than the target account that are logged in and have the same target annotation object corresponding to the target account.

[0104] Step S104: determining the images to be pushed to each target account based on the images to be annotated of each target account;

[0105] After retrieving the image to be annotated corresponding to each target account, it is hoped to push this image to the electronic device of the target account corresponding to the image so that the staff corresponding to the target account can perform subsequent annotation. Therefore, the image to be pushed of the target account is determined among the images to be annotated of the target account.

[0106] Optionally, in a specific implementation, such as Figure 2a As shown, the above step S104: determining the image to be pushed for each target account according to the image to be annotated of each target account, may include the following step S201:

[0107] Step S201: Determine the image to be annotated of each target account as the image to be pushed to the target account;

[0108] Among them, the image to be pushed determined as the target account is the complete image to be annotated of the target account; wherein, the image to be annotated not only includes the target annotation object corresponding to the target account, but also includes other annotation objects except the target annotation object, and, since the first correspondence relationship restricts the authority of the account, the staff of the account can only perform image annotation on the target annotation object determined by the account based on the first correspondence relationship, and the scope of operation is limited to the target annotation object.

[0109] Optional, such as Figure 2b As shown, the above step S104 may include the following step S202:

[0110] Step S202: determining the sub-image of the target account's designated annotation object in the target account's to-be-annotated image as the target account's to-be-pushed image.

[0111] The sub-images of the target annotated object may be obtained by intercepting a preset image before retrieving the image to be annotated for each target account, and then intercepting the preset image into multiple sub-images according to the annotated object.

[0112] The sub-image of the target marked object can also be obtained by capturing all preset images containing the target marked object after the target login account determines its corresponding target marked object, and only retaining the image containing the target marked object corresponding to the target account as the sub-image.

[0113] Based on this, it can be ensured that the annotated range of the image to be annotated obtained by the terminal device logged in with the target account is only the target annotated object corresponding to the target account. Therefore, it can effectively avoid the annotated objects that do not meet the first correspondence with the target annotated account from being unannotated, thereby improving the accuracy of image annotation.

[0114] Step S105: Pushing the image to be pushed of each target account to the target terminal device, so that the target terminal device displays the received image to be pushed;

[0115] Step S106: For each target account, obtain the annotation information of the image to be pushed to the target account;

[0116] The above-mentioned annotation information is sent by the above-mentioned target terminal device, and is annotation information about a designated annotation object in the image to be pushed of the target account. The above-mentioned designated annotation object is a target annotation object corresponding to the target account.

[0117] After determining the image to be pushed for each target account, in order to enable the staff to annotate the target annotation object, the image to be pushed can be pushed to the target terminal logged in by the account used by the staff. When the target terminal device receives the image to be pushed, the image to be pushed can be displayed in the target terminal device. The annotation result obtained after the staff annotates the annotation object in the image to be pushed is the annotation information of the annotation object.

[0118] Among them, the electronic device can directly push all the images to be pushed corresponding to each target account to the terminal device logged in with the target account; it can also divide the images to be pushed into multiple batches according to the specified quantity and push them to the terminal device logged in with the target account; for the sake of clarity, step S105 will be explained in detail later.

[0119] Optionally, in a specific implementation, the above step S105: pushing the image to be pushed of each target account to the above target terminal device may include the following step B1:

[0120] Step B1: For each target account, a specified number of remaining images are pushed to the target terminal device until the target account terminates login or no remaining images exist;

[0121] The so-called remaining images are: the images to be pushed that have not been pushed to the target terminal device of the target account;

[0122] In this specific implementation, before the first push, the so-called remaining images are all the images to be pushed that are determined to be pushed to the target terminal device.

[0123] When pushing the images to be annotated corresponding to each target account to the target terminal device, a specified number of remaining images can be pushed to the target terminal device first. After the target account completes the image annotation of the specified number of remaining images, the electronic device can determine the number of remaining images so as to continue pushing the specified number of remaining images to the target terminal device.

[0124] Among them, when the number of remaining images is greater than the specified number, the specified number of remaining images will still be pushed to the target terminal device; when there are remaining images and the number of remaining images is less than the specified number of remaining images, all remaining images will be pushed to the target terminal device and the push will be stopped after this push; when there are no remaining images, the remaining images will also stop being pushed to the target terminal device.

[0125] Moreover, in the above process, when it is detected that the target account has logged out, the electronic device stops pushing images to the target terminal device. For the remaining images in the electronic device that have not been pushed, the remaining images can be pushed again when the target account logs in next time. Alternatively, when it is detected that other accounts other than the above target account that have a first corresponding relationship with the marked objects in the remaining images log in to the terminal device, the above remaining images that have not been pushed can be pushed to the new account.

[0126] That is, in the electronic device, the number of remaining images can be judged, and the number of remaining images to be pushed to the terminal device in the next batch can be determined according to the judgment result.

[0127] For example, if a target account has 100 images to be annotated, then for this target account, the remaining 30 images can be pushed to the target terminal device. When the target account completes the image annotation of these 30 remaining images, the remaining 30 annotated images will continue to be pushed to the target account until there are 10 remaining images that have not been pushed. At this time, all 10 images will be pushed to the target terminal device.

[0128] Based on this, a certain number of images to be pushed can be pushed to the target terminal device to avoid the terminal device from freezing due to pushing all the images to be annotated to the terminal device at one time, which in turn leads to reduced image annotation efficiency or image loss.

[0129] Optionally, the above step S105 may further include the following step B2:

[0130] Step B2: For each target account, after obtaining the annotation information of the preset number of images to be pushed for the target account, push the above preset number of remaining images to the target terminal device until the target account terminates login, or there are no remaining images.

[0131] In this specific implementation method, when pushing the images to be annotated corresponding to each target account to the target terminal device, it is also possible to push the annotation information of a preset number of images to be pushed based on the target account. After the target account completes the image annotation of a preset number of remaining images, the preset number of remaining images will be pushed to the terminal device logged in with the target account until the target account terminates the login, or the number of remaining images is less than the preset number of images.

[0132] For example, if a target account corresponds to 50 images to be annotated, then for the target account, 10 images to be annotated will be pushed to the target account for the first time. After obtaining the annotation information of the two images to be pushed by the target account, the remaining two images will be pushed to the target terminal device, thereby ensuring that the number of remaining images in the terminal device of the target account always remains at 10, until the target account terminates login or the number of remaining images is less than 2, at which time all the remaining images will be pushed to the target terminal device.

[0133] In addition, for the remaining images of different target accounts, the number of remaining images pushed to the terminal device can be the same, or it can be set to different preset numbers according to the difficulty of the target annotation objects corresponding to different remaining images, which is reasonable.

[0134] Based on this, a certain number of images to be pushed can be pushed to the target terminal device to avoid the terminal device from freezing due to pushing all the images to be annotated to the terminal device at one time, which in turn leads to reduced image annotation efficiency or image loss.

[0135] As can be seen from the above, by applying the technical solution provided in the embodiment of the present application, the workgroups to which the multiple target accounts logged in in parallel belong are first determined, and then for each target account, the annotation object corresponding to the workgroup to which the target account belongs is determined based on the first correspondence between the workgroup and the annotation object, and based on the determined annotation object, the annotation task of the target account is generated, and then for each target account, the designated annotation object of the target account is determined, and based on the designated annotation object, the image to be annotated of the target account is retrieved, and then based on the image to be annotated of each target account, the image to be pushed of each target account is determined, and then the image to be pushed of each target account is pushed to the target terminal device logged in with the target account, so that the target terminal device displays the received image to be pushed, and finally, for each target account, the annotation information of the image to be pushed of the target account is obtained.

[0136] Optionally, in a specific implementation, the above step S101: for each target account, determining the marked object corresponding to the workgroup to which the target account belongs based on the first correspondence between the workgroup and the marked object, may include the following step C1:

[0137] Step C1: For each target account, search the pre-acquired annotation scheme for the workgroup to which the target account belongs, and search the annotation scheme for the annotation object corresponding to the workgroup to which the target account belongs;

[0138] The marking scheme includes: a first correspondence between a work group and a marked object, and a subordinate relationship between an account and a work group.

[0139] Optionally, the above-mentioned annotation scheme further includes: a second correspondence between the annotation object, the candidate annotation content and the annotation method.

[0140] Based on this, optionally, in a specific implementation, before the above step S104: pushing the image to be annotated corresponding to each target account to the target terminal device, the above method further includes step D1:

[0141] Step D1: Determine candidate annotation content and annotation method for the designated annotation object of each target account according to the second corresponding relationship in the above annotation scheme, as the candidate annotation content and annotation method for the target account;

[0142] The so-called second corresponding relationship is pre-set in the above-mentioned marking scheme;

[0143] After establishing the second corresponding relationship, the staff can perform image annotation on the corresponding target annotation object according to the determined candidate annotation content and annotation method.

[0144] After determining the annotation objects of the account used by the staff member, for each annotation object, candidate annotation content and annotation method of each annotation object can be determined based on the above second corresponding relationship.

[0145] Optionally, the above step S104: pushing the image to be pushed of each target account to the above target terminal device so that the above target terminal device displays the received image to be pushed may include the following step D2:

[0146] Step D2: Pushing the image to be pushed, which carries the candidate annotation content and annotation method of each target account, to the target terminal device, so that the target terminal device displays the received image to be pushed, candidate annotation content, and annotation method;

[0147] The above-mentioned candidate annotation content and annotation method can be that after the preset annotation scheme determines the annotation object of the image to be annotated, appropriate candidate annotation content and annotation method are configured according to the type of annotation object and annotation requirements, and the configured candidate annotation content and annotation method are added to the annotation object of the image to be annotated.

[0148] In this way, the candidate annotation content and annotation method of the target annotation object can be directly displayed in the image to be annotated of the target account, which is convenient for intuitive annotation of the image to be annotated, thereby improving the efficiency of image annotation; in addition, displaying the candidate annotation content and annotation method of the target annotation object in the image to be annotated of the target account also facilitates subsequent review and integration of annotation information.

[0149] Optionally, the above step S105: for each target account, obtaining the annotation information of the image to be pushed of the target account may include the following step D3:

[0150] Step D3: For each target account, obtain the annotation information of the target account's image to be pushed selected from the candidate annotation content of the target account according to the annotation method of the target account;

[0151] After the target account obtains the corresponding image to be pushed, the annotation information of the image to be pushed is selected from the candidate annotation content of the target account, and the image to be pushed is annotated according to the annotation information.

[0152] Based on this, the annotation of the image to be pushed is completed according to the annotation information selected from the candidate annotation content of the target account, which can effectively avoid slow annotation speed and annotation errors caused by unclear annotation information, thereby improving the accuracy and efficiency of image annotation.

[0153] Optionally, in a specific implementation, such as Figure 3 As shown, the image annotation method provided in the embodiment of the present application further includes the following step S301:

[0154] Step S301: In the first correspondence, add a new entry about the correspondence between the workgroup and the marked object, and / or delete the current entry about the object relationship between the workgroup and the marked object, and return to the step of determining multiple target accounts for parallel login.

[0155] After completing a labeling scheme to annotate an image, you may need to perform new image annotations, also known as supplementary annotations. Supplementary annotation requires determining new annotation objects and establishing a primary correspondence between workgroups and annotation objects in the labeling scheme. This allows staff to log in to their account, identify the workgroup to which their account belongs, and then determine the new annotation objects based on this new primary correspondence to supplement the target image.

[0156] Similarly, after completing the annotation of all images of some labeled objects, these labeled objects can be deleted to avoid wasting resources and space.

[0157] In this way, when there are newly added annotation objects in the image to be annotated, the first corresponding relationship can be updated in a timely manner, so as to facilitate subsequent image annotation of the newly added annotation objects.

[0158] Optionally, in a specific implementation, the image annotation method provided in the embodiment of the present application further includes the following step E1:

[0159] Step E1: In the annotation scheme, add a new entry about the affiliation between the account and the workgroup, and / or delete the current entry about the affiliation between the account and the workgroup, and return to the step of determining multiple target accounts for parallel login.

[0160] When the number of annotation objects increases, you can add new entries for accounts and workgroups to the annotation plan. This allows you to add more staff members to use accounts after adding new annotation objects, thereby completing the annotation of the annotation objects more efficiently. In addition, when some accounts are no longer used for image annotation work, they can be deleted to avoid using accounts that cannot perform annotation work for image annotation.

[0161] Optionally, in a specific implementation, such as Figure 4 As shown, the image annotation method provided in the embodiment of the present application further includes the following steps S401-S402:

[0162] Step S401: determining target annotation information of each target image that has completed information annotation in the preset image, and integrating the target annotation information of each target image to obtain each target image carrying target integration information;

[0163] In order to use sample images that have been annotated for pattern model training, it is necessary to integrate the annotation information of the images that have completed information annotation to facilitate subsequent sample training; wherein, the above-mentioned annotation information may only include the annotation information of the initially annotated image, or only include the annotation information of the supplemented annotated image, or may include both the initial annotation and the supplemented annotation.

[0164] Step S402: Send each target image carrying the target integration information to a designated device.

[0165] The designated device may be a terminal device logged in by the account for marking the target image, or may be other terminal devices or electronic devices designated according to needs.

[0166] In this specific implementation, all target annotation information of each target image that has completed image annotation may be integrated; or part of the required target annotation information may be selected for integration according to actual needs, thereby obtaining each target image carrying target integration information.

[0167] Based on this, after obtaining the above-mentioned target images carrying target integration information, they are sent to the designated device to facilitate subsequent inspection of the target images based on the target integration information carried by the target images, so that the target images that need to be modified in image annotation can be discovered in time, and the preset annotation scheme can be updated, thereby improving the accuracy and efficiency of image annotation.

[0168] Corresponding to the image annotation method provided in the above embodiment of the present application, the embodiment of the present application also provides an image annotation device such as Figure 5 As shown, the device includes:

[0169] A workgroup determination module 501 is used to determine the workgroups to which multiple target accounts logged in in parallel belong respectively;

[0170] The task generation module 502 is configured to determine, for each target account, the annotation object corresponding to the workgroup to which the target account belongs based on the first correspondence between the workgroup and the annotation object, and generate an annotation task for the target account based on the determined annotation object, wherein the annotation task for each target account includes: the annotation object corresponding to the workgroup to which the target account belongs;

[0171] The image retrieval module 503 is configured to determine, for each target account, a designated annotation object for the target account, and retrieve images to be annotated for the target account based on the designated annotation object; wherein the designated annotation object is the object indicated by the received object selection instruction in the annotation task for the target account; the image to be annotated is an image of a terminal device in which the designated annotation object contained in a preset image is not annotated and has not been pushed outside a target terminal device, and the object indicated by the received object selection instruction is the designated annotation object; the target terminal device is the terminal device logged in with the target account;

[0172] An image determination module 504 is configured to determine an image to be pushed to each target account based on the image to be annotated of each target account;

[0173] An image pushing module 505 is configured to push the image to be pushed of each target account to the target terminal device, so that the target terminal device displays the received image to be pushed;

[0174] The information acquisition module 506 is used to obtain the annotation information of the image to be pushed of each target account; wherein the annotation information is sent by the target terminal device and is the annotation information of the specified annotation object in the image to be pushed of the target account.

[0175] As can be seen from the above, by applying the technical solution provided in the embodiment of the present application, the workgroups to which the multiple target accounts logged in in parallel belong are first determined, and then for each target account, the annotation object corresponding to the workgroup to which the target account belongs is determined based on the first correspondence between the workgroup and the annotation object, and based on the determined annotation object, the annotation task of the target account is generated, and then for each target account, the designated annotation object of the target account is determined, and based on the designated annotation object, the image to be annotated of the target account is retrieved, and then based on the image to be annotated of each target account, the image to be pushed of each target account is determined, and then the image to be pushed of each target account is pushed to the target terminal device logged in with the target account, so that the terminal device displays the received image to be pushed, and finally, for each target account, the annotation information of the image to be pushed of the target account is obtained.

[0176] Based on this, for multiple objects that need to be labeled, the objects to be labeled are assigned to different workgroups, thereby establishing a first correspondence between the workgroup and the labeled object. Multiple accounts for image annotation are assigned to the workgroup. After the staff member logs in to the account, the workgroup to which the account belongs is determined, and the labeled object is determined according to the first correspondence, thereby completing the labeling of the labeled object. In this way, when labeling, different staff members can log in to the account in parallel, so that the corresponding images to be labeled can be labeled in parallel. Since multiple accounts are logged in in parallel, multiple labeled objects of multiple images can be labeled at the same time, so the efficiency of image labeling can be guaranteed. Moreover, since different staff members' accounts are assigned different labeled objects, the staff member assigned to each labeled object does not need to pay attention to all the labeled objects. Therefore, it is not easy for the staff member to make mistakes when labeling the image. While ensuring the efficiency of image labeling, the accuracy of image labeling can be improved, thereby improving the accuracy of image samples, and further improving the accuracy of subsequent image model training.

[0177] Optionally, in a specific implementation, the image determination module 504 is specifically configured to:

[0178] Determine the image to be annotated for each target account as the image to be pushed to the target account; or

[0179] A sub-image including the target annotation object corresponding to the target account in the image to be annotated of each target account is determined as the image to be pushed of the target account.

[0180] Optionally, in a specific implementation, the image pushing module 505 is specifically configured to:

[0181] For each target account, a specified number of remaining images are pushed to the terminal device; wherein the remaining images are: images to be pushed of the target account that have not been pushed to the target terminal device;

[0182] The device further comprises:

[0183] The remaining image pushing module is used to trigger the image pushing module for each target account after obtaining the annotation information of all the images to be pushed that have been pushed by the target account, until the target account terminates login, or there are no remaining images; or, for each target account, after obtaining the annotation information of a preset number of images to be pushed that have been pushed by the target account, push the preset number of remaining images to the target terminal device until the target account terminates login, or there are no remaining images.

[0184] Optionally, in a specific implementation, the task generation module 502 is specifically configured to:

[0185] For each target account, search the pre-acquired annotation scheme for the workgroup to which the target account belongs, and search the annotation scheme for the annotation object corresponding to the workgroup to which the target account belongs;

[0186] The marking scheme includes: a first corresponding relationship between a work group and a marked object, and a subordinate relationship between an account and a work group.

[0187] Optionally, in a specific implementation, the annotation scheme further includes: a second correspondence between the annotation object, the candidate annotation content, and the annotation method;

[0188] The device further comprises:

[0189] a content and method determination module configured to determine, before pushing the to-be-annotated image corresponding to each target account to the target terminal device, candidate annotation content and annotation method for the designated annotation object of each target account based on the second corresponding relationship in the annotation scheme, as the candidate annotation content and annotation method for the target account;

[0190] The image push module 505 is specifically used for:

[0191] Pushing the image to be pushed, which carries the candidate annotation content and annotation method of each target account, to the target terminal device, so that the target terminal device displays the received image to be pushed, candidate annotation content, and annotation method;

[0192] The information acquisition module 506 is specifically used to:

[0193] For each target account, the annotation information of the target account's to-be-pushed image selected from the candidate annotation contents of the target account according to the annotation method of the target account is obtained.

[0194] Optionally, in a specific implementation, the device further includes:

[0195] The first module for adding and deleting entries is used to add new entries about the correspondence between the workgroup and the marked object in the first correspondence, and / or delete the current entries about the object relationship between the workgroup and the marked object, and return to the step of determining multiple target accounts for parallel login.

[0196] Optionally, in a specific implementation, the device further includes:

[0197] The second module for adding and deleting entries is used to add new entries about the subordinate relationship between the account and the workgroup in the annotation scheme, and / or delete the current entries about the subordinate relationship between the account and the workgroup, and return to the step of determining multiple target accounts for parallel login.

[0198] Optionally, in a specific implementation, the device further includes:

[0199] An information integration module is used to determine target annotation information of each target image that has completed information annotation in the preset image, and integrate the target annotation information of each target image to obtain each target image carrying target integration information;

[0200] The image sending module is used to send each target image carrying target integration information to a designated device.

[0201] It should be noted that the images to be annotated in this embodiment are all from public datasets.

[0202] The present application also provides an electronic device, such as Figure 6 As shown, including:

[0203] Memory 601, used for storing computer programs;

[0204] The processor 602 is configured to implement any of the image annotation methods provided in the above-mentioned embodiments of the present application when executing the program stored in the memory 601 .

[0205] Optionally, the communication bus mentioned in the electronic device may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. Such a communication bus may be divided into an address bus, a data bus, a control bus, and the like. For ease of illustration, the figure shows only one thick line, but this does not imply that there is only one bus or only one type of bus.

[0206] The communication interface is used for communication between the above electronic device and other devices.

[0207] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage. Alternatively, the memory may be at least one storage device located away from the processor.

[0208] The above-mentioned 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 gate or transistor logic devices, and discrete hardware components.

[0209] Furthermore, the electronic device may further include a communication bus and / or a communication interface, and the processor 602, the communication interface, and the memory 601 communicate with each other via the communication bus.

[0210] In another embodiment provided by the present application, a computer-readable storage medium is further provided, wherein a computer program is stored in the computer-readable storage medium. When the computer program is executed by a processor, the steps of any of the above-mentioned image annotation methods are implemented.

[0211] In another embodiment provided by the present application, a computer program product including instructions is further provided, which, when executed on a computer, enables the computer to execute any one of the image annotation methods in the above embodiments.

[0212] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When software is used for implementation, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website, computer, server or data center to another website, computer, server or data center via a wired (e.g., coaxial cable, optical fiber, digital staff line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrations. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), etc.

[0213] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0214] Each embodiment in this specification is described in a related manner. Similar parts between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the system embodiment is generally similar to the method embodiment, so the description is relatively simple. For related parts, refer to the description of the method embodiment.

[0215] The above description is only a preferred embodiment of the present application and is not intended to limit the scope of protection of the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application are included in the scope of protection of the present application.

Claims

1. An image annotation method, characterized in that: The method comprises: Determine the workgroups to which multiple target accounts logged in in parallel belong; For each target account, determining the annotation object corresponding to the workgroup to which the target account belongs based on the first correspondence between the workgroup and the annotation object, and generating a annotation task for the target account based on the determined annotation object; wherein the annotation task for each target account includes: the annotation object corresponding to the workgroup to which the target account belongs; For each target account, a designated annotation object for the target account is determined, and based on the designated annotation object, images to be annotated for the target account are retrieved; wherein the designated annotation object is the object indicated by the received object selection instruction in the annotation task for the target account, and the image to be annotated is an image of a terminal device in which the designated annotation object contained in a preset image is not annotated and has not been pushed outside a target terminal device, and the object indicated by the received object selection instruction is the designated annotation object; the target terminal device is the terminal device on which the target account is logged in; Determine the images to be pushed to each target account based on the images to be annotated for each target account; Pushing the image to be pushed of each target account to the target terminal device, so that the target terminal device displays the received image to be pushed; For each target account, obtain the annotation information of the image to be pushed of the target account; wherein the annotation information is sent by the target terminal device and is the annotation information of the specified annotation object in the image to be pushed of the target account.

2. The method according to claim 1, characterized in that The step of determining the image to be pushed for each target account according to the image to be annotated for each target account includes: Determine the image to be annotated for each target account as the image to be pushed to the target account; or A sub-image including the designated annotation object of the target account in the image to be annotated of each target account is determined as the image to be pushed of the target account.

3. The method according to claim 1, characterized in that The step of pushing the image to be pushed of each target account to the target terminal device includes: For each target account, a specified number of remaining images are pushed to the target terminal device; wherein the remaining images are: images to be pushed of the target account that have not been pushed to the target terminal device; The method further comprises: For each target account, after obtaining the annotation information of all images to be pushed that have been pushed by the target account, returning to the step of pushing the specified number of remaining images to the target terminal device until the target account terminates login or there are no remaining images; or, For each target account, after obtaining the annotation information of the preset number of images to be pushed that have been pushed to the target account, the preset number of remaining images will be pushed to the target terminal device until the target account terminates login, or there are no remaining images.

4. The method according to claim 1, wherein The step of determining, for each target account, the marked object corresponding to the workgroup to which the target account belongs based on the first correspondence between the workgroup and the marked object includes: For each target account, search the pre-acquired annotation scheme for the workgroup to which the target account belongs, and search the annotation scheme for the annotation object corresponding to the workgroup to which the target account belongs; The marking scheme includes: a first corresponding relationship between a work group and a marked object, and a subordinate relationship between an account and a work group.

5. The method according to claim 4, characterized in that The annotation scheme further includes: a second correspondence between the annotation object, the candidate annotation content, and the annotation method; Before pushing the to-be-annotated image corresponding to each target account to the target terminal device, the method further includes: Determining, based on the second corresponding relationship in the annotation scheme, candidate annotation content and annotation method for the designated annotation object of each target account as the candidate annotation content and annotation method for the target account; The step of pushing the image to be pushed of each target account to the target terminal device so that the target terminal device displays the received image to be pushed includes: Pushing the image to be pushed, which carries the candidate annotation content and annotation method of each target account, to the target terminal device, so that the target terminal device displays the received image to be pushed, candidate annotation content, and annotation method; The step of obtaining, for each target account, the annotation information of the image to be pushed of the target account includes: For each target account, the annotation information of the target account's to-be-pushed image selected from the candidate annotation contents of the target account according to the annotation method of the target account is obtained.

6. The method according to claim 1, characterized in that The method further comprises: In the first correspondence, a new entry regarding the correspondence between the workgroup and the marked object is added, and / or the current entry regarding the object relationship between the workgroup and the marked object is deleted, and the step of determining multiple target accounts for parallel login is returned.

7. The method according to claim 4, characterized in that The method further comprises: In the marking scheme, a new entry about the affiliation between the account and the workgroup is added, and / or a current entry about the affiliation between the account and the workgroup is deleted, and the step of determining multiple target accounts for parallel login is returned.

8. The method according to any one of claims 1 to 7, characterized in that The method further comprises: Determining target annotation information of each target image that has completed information annotation in the preset image, and integrating the target annotation information of each target image to obtain each target image carrying target integration information; Each target image carrying target integration information is sent to the designated device.

9. An image annotation device, characterized in that: The device comprises: A workgroup determination module is used to determine the workgroups to which multiple target accounts logged in in parallel belong; a task generation module configured to determine, for each target account, the labeled object corresponding to the workgroup to which the target account belongs based on the first correspondence between the workgroup and the labeled object, and generate a labeling task for the target account based on the determined labeled object, wherein the labeling task for each target account includes: the labeled object corresponding to the workgroup to which the target account belongs; An image retrieval module is configured to determine, for each target account, a designated annotation object for the target account, and retrieve images to be annotated for the target account based on the designated annotation object; wherein the designated annotation object is the object indicated by the received object selection instruction in the annotation task for the target account; the image to be annotated is an image of a terminal device in which the designated annotation object contained in a preset image is not annotated and has not been pushed outside a target terminal device, and the object indicated by the received object selection instruction is the designated annotation object; the target terminal device is the terminal device on which the target account is logged in; An image determination module, configured to determine the image to be pushed to each target account based on the image to be annotated of each target account; An image pushing module, configured to push the image to be pushed of each target account to the target terminal device, so that the target terminal device displays the received image to be pushed; The information acquisition module is used to obtain the annotation information of the image to be pushed of each target account; wherein the annotation information is sent by the target terminal device and is the annotation information of the specified annotation object in the image to be pushed of the target account.

10. The device according to claim 9, characterized in that The image determination module is specifically used for: Determine the image to be annotated for each target account as the image to be pushed to the target account; or Determine the sub-image of the target annotation object corresponding to each target account in the image to be annotated of the target account as the image to be pushed of the target account; and / or, The image push module is specifically used for: For each target account, a specified number of remaining images are pushed to the target terminal device; wherein the remaining images are: images to be pushed of the target account that have not been pushed to the target terminal device; The device further comprises: The remaining image push module is configured to trigger the image push module for each target account after obtaining the annotation information of all images to be pushed that have been pushed by the target account, until the target account terminates login, or no remaining images exist; or, for each target account, after obtaining the annotation information of a preset number of images to be pushed that have been pushed by the target account, push the preset number of remaining images to the target terminal device, until the target account terminates login, or no remaining images exist; and / or, The task generation module is specifically used to: For each target account, search the pre-acquired annotation scheme for the workgroup to which the target account belongs, and search the annotation scheme for the annotation object corresponding to the workgroup to which the target account belongs; The marking scheme includes: a first correspondence between a work group and a marked object, and a subordinate relationship between an account and a work group; and / or, The annotation scheme further includes: a second correspondence between the annotation object, the candidate annotation content, and the annotation method; The device further comprises: a content and method determination module configured to determine, before pushing the to-be-annotated image corresponding to each target account to the target terminal device, candidate annotation content and annotation method for the designated annotation object of each target account based on the second corresponding relationship in the annotation scheme, as the candidate annotation content and annotation method for the target account; The image push module is specifically used for: Pushing the image to be pushed, which carries the candidate annotation content and annotation method of each target account, to the target terminal device, so that the target terminal device displays the received image to be pushed, candidate annotation content, and annotation method; The information acquisition module is specifically used for: For each target account, obtain the annotation information of the target account's image to be pushed selected from the candidate annotation content of the target account according to the annotation method of the target account; and / or, The device further comprises: a first entry addition and deletion module, configured to add a new entry regarding the correspondence between the workgroup and the marked object in the first correspondence, and / or delete a current entry regarding the object relationship between the workgroup and the marked object, and return to the step of determining the multiple target accounts for parallel login; and / or, The device further comprises: a second entry addition and deletion module configured to add a new entry regarding the affiliation between the account and the workgroup in the annotation scheme, and / or delete a current entry regarding the affiliation between the account and the workgroup, and return to the step of determining the multiple target accounts for concurrent login; and / or, The device further comprises: An information integration module is used to determine target annotation information of each target image that has completed information annotation in the preset image, and integrate the target annotation information of each target image to obtain each target image carrying target integration information; The image sending module is used to send each target image carrying target integration information to a designated device.

11. An electronic device, characterized in that: include: Memory for storing computer programs; A processor, configured to implement the method according to any one of claims 1 to 8 when executing a program stored in a memory.

12. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images