CN111324905A - Image data labeling method and device, computer equipment and storage medium - Google Patents

Abstract

The present application relates to the field of data security, and in particular, to a method and an apparatus for labeling image data, a computer device, and a storage medium. The method comprises the following steps: receiving a task marking request sent by a user terminal, wherein the task marking request carries a task identifier; inquiring a storage address and a first key of an image file corresponding to the task identifier from a task list stored in a local database; sending the first key to a storage address in a cloud, wherein the cloud is used for verifying the first key, and returning the desensitized image file after successful verification; sending the desensitized image file to a user terminal so that the user terminal can analyze the image file to obtain pixel information and desensitization information and display the obtained pixel information and desensitization information; and receiving the structured data after the pixel information is labeled by the user terminal, and uploading the structured data to the cloud end so as to perform associated storage with the image file stored in the cloud end. By adopting the method, the data security can be improved.

Description

Image data labeling method, device, computer equipment and storage medium

technical field

The present application relates to the field of big data technology, and in particular, to an image data labeling method, device, computer equipment and storage medium.

Background technique

Data annotation is an important part of the field of artificial intelligence. It is an essential stage for building algorithm models and improving algorithm accuracy. For data annotation, the annotation platform provides such a basic annotation function to provide data annotation services for algorithms.

For image annotation only, traditional annotation tool products are generally implemented based on the client, because the client has relatively complete functions and a good user experience, especially the performance of loading local files has advantages, but it also faces data security aspects. The problem: when a large number of users use these client products for data annotation, since they need to load the data locally, there is a risk of leaking the annotation data.

SUMMARY OF THE INVENTION

Based on this, it is necessary to provide an image data labeling method, device, computer equipment and storage medium that can ensure data security in view of the above technical problems.

A method for labeling image data, the method comprising:

receiving a task labeling request sent by the user terminal, where the task labeling request carries a task identifier;

From the task list stored in the local database, query the storage address and first key of the image file corresponding to the task identifier;

Sending the first key to the storage address in the cloud, where the cloud is used to verify the first key, and after successful verification, returns the desensitized image file in the form of a binary stream;

Receive the desensitized image file returned by the cloud in a binary stream, and send the desensitized image file to the user terminal in a binary stream, so that the user terminal parses the image file to obtain pixel information and desensitization information, and display the obtained pixel information and desensitization information;

Receive the structured data marked by the user terminal on the pixel information, and upload the structured data to the cloud for associated storage with the image files stored in the cloud.

In one embodiment, before receiving the task labeling request sent by the user terminal, the method further includes:

receiving a labeling request sent by the user terminal, and querying the task list stored in the first database according to the labeling request to obtain tasks whose status is to be labelled;

Sending the task whose status is the status to be marked to the user terminal, and the task whose status is the status to be marked is used for selection by the user terminal.

In one embodiment, after uploading the structured data to the cloud for associated storage with the image files stored in the cloud, the method further includes:

Receive the second key corresponding to the structured data returned by the cloud, store the second key in the task corresponding to the task list, and store the corresponding task in the task list. The status is changed to pending review;

Receive the audit request sent by the audit terminal, and query the tasks in the task list whose status is to be audited according to the audit request;

Obtaining the task whose status is to be reviewed and returning it to the review terminal, and receiving the selected task returned by the review terminal;

Query the second key corresponding to the selected task, and return the second key to the audit terminal, where the second key is used for the audit terminal to download the structured data from the cloud And conduct audit, and store the audit result to the local database.

In one embodiment, after storing the audit result in the local database, the method further includes:

Send the task and the corresponding second key corresponding to the structured data that the audit result is passed to the research and development platform for storage, and the task and the second key stored in the research and development platform are used for the research and development terminal to query the corresponding task. and download the structured data corresponding to the audited task from the cloud through the second key.

In one embodiment, the querying the task address and the first key corresponding to the task identifier from the task list stored in the local database includes:

The processor is selected for execution through a load balancing algorithm, and the task address and the first key corresponding to the task identifier are queried from the task list stored in the local database.

An image data labeling device, the device comprising:

a first receiving module, configured to receive a task labeling request sent by a user terminal, where the task labeling request carries a task identifier;

a first query module, configured to query the storage address and the first key of the image file corresponding to the task identifier from the task list stored in the local database;

The first sending module is used to send the first key to the storage address in the cloud, and the cloud is used to verify the first key, and after the verification is successful, it returns a binary stream to the address. post-sensitization image files;

The second receiving module is configured to receive the desensitized image file returned by the cloud in the form of binary stream, and send the desensitized image file to the user terminal in the form of binary stream, so that the user terminal can parse the image file. The image file obtains pixel information and desensitization information, and displays the obtained pixel information and desensitization information;

A second sending module, configured to receive structured data marked by the user terminal on the pixel information, and upload the structured data to the cloud to associate with the image files stored in the cloud storage.

In one embodiment, the apparatus further includes:

a third receiving module, configured to receive a labeling request sent by the user terminal, and query the task list stored in the first database according to the labeling request to obtain tasks whose status is to be labelled;

The third sending module is configured to send the task whose status is to be marked to the user terminal, and the task whose status is to be marked is used for selection by the user terminal.

In one embodiment, the apparatus further includes:

a fourth receiving module, configured to receive the second key corresponding to the structured data returned by the cloud, store the second key in the corresponding task in the task list, and store the second key in the task list. The status of the corresponding task in the task list is changed to pending review;

A second query module, configured to receive an audit request sent by the audit terminal, and query the tasks in the task list that are in a state to be audited according to the audit request;

a selection module, configured to obtain the task whose status is to be reviewed and return it to the review terminal, and receive the selected task returned by the review terminal;

an audit module, configured to query the second key corresponding to the selected task, and return the second key to the audit terminal, where the second key is used for the audit terminal to download from the cloud The structured data is audited, and the audit results are stored in the local database.

A computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of any one of the methods described above when the processor executes the computer program.

A computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, implements the steps of any of the methods described above.

The above-mentioned image data labeling method, device, computer equipment and storage medium are introduced into the cloud, and when the user labels the image data, only the image file is returned as a binary stream, including the pixel information and the desensitized information after desensitization. Ensure the security of image files. And in the whole process, the user terminal is isolated from the image file throughout the whole process, and the user cannot obtain the user information in the image file, which ensures the security of the image file. In addition, the marked data is uploaded to the cloud, avoiding the risk of data leakage.

Description of drawings

1 is an application scenario diagram of a method for labeling image data in one embodiment;

2 is a schematic flowchart of a method for labeling image data in one embodiment;

FIG. 3 is an interface diagram of an annotation page in one embodiment;

4 is a structural block diagram of an image data labeling apparatus in one embodiment;

FIG. 5 is a diagram of the internal structure of a computer device in one embodiment.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present application more clearly understood, the present application will be described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, but not to limit the present application.

The image data labeling method provided in this application can be applied to the application environment shown in FIG. 1 . The user terminal 102 communicates with the server 104 through the network, and the server 104 communicates with the cloud. The user terminal 102 can download the task list from the server 104, select a task to execute, and after selecting the task, send the task labeling request to the server 104, and the server 104 searches the task list stored in the local database 108 for the corresponding task identifier. The storage address and first key of the image file, and send the first key to the storage address in the cloud, so that the cloud can verify the first key, and after the verification is successful, return the desensitized image in the form of a binary stream. image file. Therefore, the server 104 sends the desensitized image file to the user terminal 102 in a binary stream mode, so that the user terminal 102 analyzes the image file to obtain pixel information and desensitization information, and displays the obtained pixel information and desensitization information; The terminal 102 annotates the pixel information to generate structured data, and returns the annotated structured data to the server 104, and the server 104 uploads the structured data to the cloud for associated storage with the image files stored in the cloud. In addition, an audit terminal 106 is included for auditing the marked structured data. The terminal 102 can be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers and portable wearable devices, and the server 104 can be implemented by an independent server or a server cluster composed of multiple servers.

In one embodiment, as shown in FIG. 2, a method for labeling image data is provided, and the method is applied to the server in FIG. 1 as an example for description, including the following steps:

S202: Receive a task labeling request sent by the user terminal, where the task labeling request carries a task identifier.

Specifically, the task labeling request is generated by the user terminal, which carries the task that the user terminal wants to perform labeling operation, and each task stored in the task list has a unique identifier. In practical applications, the task labeling request is generated after the user terminal receives a user's instruction for selecting tasks displayed on the terminal. That is, the user terminal displays tasks in the form of a list, etc. The user selects the task to be processed by clicking operations, etc., so that the terminal can query the task ID of the task selected by the user, and generate a task annotation request according to the task ID, so that the task The callout request is sent to the server for processing.

S204: From the task list stored in the local database, query the storage address and the first key of the image file corresponding to the task identifier.

Specifically, the local database refers to the database corresponding to the annotation platform, and the local database stores a task list in which various tasks are stored, including task status, task identifier, storage address of the image file corresponding to the task, and a first key. Wait. The task status includes the status to be marked, the status to be reviewed, and the status of marked completion. The first key is the image file corresponding to the task, that is, the first key generated when the dicom file is stored in the public cloud, through which the image file stored in the public cloud can be obtained, that is, the server stores the first key. The key is uploaded to the public cloud obs, and then the public cloud obs judges whether the first key is accurate, and returns the image information if it is accurate; when the data source is an internal client, the key is the key composed of the nas path + file name and some other information , after the key is obtained during download, the corresponding nas file is found and converted into a file stream and returned. After receiving the task identifier, the server may query the task list according to the task identifier to obtain the corresponding storage address and the first key. The method of adding tasks in the task list in the local database may include: when the algorithm engineer uploads the image file dicom to the private cloud through the internal client or uploads the image file dicom to the public cloud through the image platform (the internal client is to upload the image file to the public cloud). It is placed on the nas of the medical service; the image platform uploads image files to the public cloud obs.), so that the cloud updates the file storage address in the database corresponding to the internal R&D platform according to the newly added image file dicom, and the server can Actively inquire whether there is an updated file storage address of the image file dicom in the database of the internal R&D platform, and if so, import it to the server. The import here refers to importing the file storage path, that is, the above storage address, into the server, not importing The entire image file dicom. The above image data is stored in the cloud, which strengthens data security management and control, ensures the security of image data and labelled results, and avoids the risk of data leakage.

The user terminal only uploads the task identifier to the server, so that the server can query the task list stored in the local database to obtain the corresponding storage address and the first key, so that the server can obtain the image file from the public cloud.

S206: Send the first key to the storage address in the cloud, and the cloud is used to verify the first key, and after successful verification, returns the desensitized image file in the form of a binary stream.

Specifically, the image file is composed of pixel information of the image and a lot of tag information, and the tag contains a lot of names and genders of patients. The server uses the first key to go to the corresponding storage address to fetch the image file, that is, the cloud first verifies the first key, and if the first key verification is successful, the cloud queries the corresponding image file dicom and identifies the image file dicom The information to be desensitized in, and then desensitized the information to be desensitized to obtain desensitization information, that is, modify the information to be desensitized to "*", and then send the desensitized information to the server in the form of a binary stream.

S208: Receive the desensitized image file returned by the cloud in a binary stream, and send the desensitized image file to the user terminal in a binary stream, so that the user terminal can analyze the image file to obtain pixel information and desensitization information, and The resulting pixel information and desensitization information are displayed.

Specifically, the server receives the image file in the form of a binary stream returned by the cloud, and then returns the received binary stream to the user terminal, so that the user terminal can parse the binary stream to obtain pixel information and desensitization information, and then analyze the pixel information and desensitization information. After the information is calculated, it is rendered and displayed, and the desensitized information is displayed. Specifically, as shown in FIG. 3, FIG. 3 is an interface diagram of an annotation page in an embodiment, in which pixel information is rendered and displayed, and desensitization information is displayed directly, thereby ensuring that the patient's name, The gender will not be obtained by the user, and the information is only saved to the cloud, and the entire labeling process will not be obtained.

The user terminal is displayed as shown in Figure 3, so that the user can delineate the characteristics of the lesion on this page and label the lesion, and then the user terminal generates structured data according to the lesion information marked by the user, and associates the structured data with the corresponding information. The task ID is returned to the server.

S210: Receive the structured data marked by the pixel information by the user terminal, and upload the structured data to the cloud, so as to be stored in association with the image files stored in the cloud.

Specifically, after the server receives the structured data returned by the user terminal, that is, json data, the structured data is uploaded to the public cloud, and the public cloud generates a second key, returns it to the server, and stores it in the local database. Therefore, when the user needs to view the marked structured data, he can query through the second key.

The above-mentioned image data labeling method is introduced into the cloud, and when the user labels the image data, only the image file is returned as a binary stream, including pixel information and desensitized information after desensitization, which can ensure the security of the image file. And in the whole process, the user terminal is isolated from the image file throughout the whole process, and the user cannot obtain the user information in the image file, which ensures the security of the image file. In addition, the marked data is uploaded to the cloud, avoiding the risk of data leakage.

In one embodiment, before receiving the task labeling request sent by the user terminal, the method further includes: receiving the labeling request sent by the user terminal, and querying the task list stored in the first database according to the labeling request to obtain a task whose status is to be labelled; The tasks whose status is the status to be marked are sent to the user terminal, and the tasks whose status is the status to be marked are used for selection by the user terminal.

Specifically, the server receives the labeling request sent by the user terminal, and according to the labeling request, queries the task list in the local database to obtain the task whose status is the status to be marked, and returns the task whose status is the status to be marked to the user terminal, So that the user terminal selects the task whose status is to be marked, and generates a task marking request according to the task identifier of the selected task.

When the user wants to label the data, he can first send a labeling request to the server. After the server receives the labeling request, it can query the local database to obtain the task whose status is to be labelled, and return the task to the user terminal. Therefore, the user terminal can display the tasks to facilitate the user's selection, and after the user makes the selection, the user terminal generates a task labeling request, and sends the task labeling request to the server.

In addition, in this embodiment, the user terminal can log in to the server through the web page at the second level to obtain the task list, instead of setting up a client, etc., the annotator only needs to register an account through the server's web page to start annotating.

In the above embodiment, the user can select tasks to be marked that he is interested in from the task list in the server for processing, thereby improving the efficiency of task processing.

In one of the embodiments, after uploading the structured data to the cloud and storing it in association with the image files stored in the cloud, the method further includes: receiving a second key corresponding to the structured data returned by the cloud, and storing the second key in the cloud. The key is stored in the corresponding task in the task list, and the state of the corresponding task in the task list is changed to the pending state; the audit request sent by the audit terminal is received, and the state in the task list is inquired according to the audit request. task; obtain the task whose status is pending review and return it to the review terminal, and receive the selected task returned by the review terminal; query the second key corresponding to the selected task, and return the second key to the review terminal, and the second key is returned to the review terminal. The key is used for the audit terminal to download structured data from the cloud and audit it, and store the audit result in the local database.

Specifically, after the marking is completed, an auditing step may also be included, that is, the auditing terminal sends an auditing request to the server, the server queries the tasks in the local database that are in the pending auditing state, and returns them to the auditing terminal, and the auditing terminal selects the tasks to be audited, and Send the task identifier of the task that needs to be reviewed to the server, the server queries the corresponding second key according to the task identifier and returns it to the audit terminal, and the audit terminal obtains the corresponding structured data from the public cloud through the second key, and After the audit is passed, the approved information is sent to the local database, and the structured data and the approved information are sent to the R&D platform, so as to store the tasks and the approved information in the database corresponding to the R&D platform.

In the above embodiment, after the labeling is completed, the review process of the labelled structured data is also included, so as to ensure the accuracy of the labelled structured data, so as to ensure the accuracy of the original data in the training process of the subsequent algorithm. to improve the accuracy of the algorithm.

In one of the embodiments, after storing the audit result in the local database, the method further includes: sending the task corresponding to the structured data whose audit result is passed and the corresponding second key to the research and development platform for storage, and storing the task in the research and development platform. The task and the second key are used for the research and development terminal to query the second key corresponding to the task, and download the structured data corresponding to the approved task from the cloud through the second key.

Specifically, the R&D terminal sends a data acquisition request to the R&D platform, the R&D platform queries the tasks in the corresponding database, obtains the second key, returns the second key to the R&D terminal, and the R&D terminal sends the cloud to the cloud through the second key Obtain the corresponding structured data, and then perform processing such as algorithm training.

In the above embodiment, after the labeling is completed, the review process of the labelled structured data is also included, so as to ensure the accuracy of the labelled structured data, so as to ensure the accuracy of the original data in the training process of the subsequent algorithm. to improve the accuracy of the algorithm.

In one of the embodiments, querying the task address and the first key corresponding to the task identifier from the task list stored in the local database includes: selecting a processor to execute through a load balancing algorithm, and obtaining the task list stored in the local database from the task list. , query the task address and the first key corresponding to the task identifier.

Specifically, this embodiment mainly emphasizes load balancing, that is, after the server receives the labeling request sent by the user terminal, it selects the corresponding processor in the labeling platform through a load balancing algorithm for processing, wherein the load balancing algorithm may be the ngnix algorithm.

In the above embodiment, the task labeling request is processed through the load balancing algorithm, which can improve the efficiency of data processing.

It should be understood that although the various steps in the flowchart of FIG. 2 are shown in sequence according to the arrows, these steps are not necessarily executed in the sequence shown by the arrows. Unless explicitly stated herein, the execution of these steps is not strictly limited to the order, and these steps may be performed in other orders. Moreover, at least a part of the steps in FIG. 2 may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed and completed at the same time, but may be executed at different times. The execution of these sub-steps or stages The sequence is also not necessarily sequential, but may be performed alternately or alternately with other steps or sub-steps of other steps or at least a portion of a phase.

In one embodiment, as shown in FIG. 4, an image data labeling apparatus is provided, including: a first receiving module 100, a first query module 200, a first sending module 300, a second receiving module 400, and a second sending module Module 500, wherein:

The first receiving module 100 is configured to receive a task labeling request sent by a user terminal, where the task labeling request carries a task identifier.

The first query module 200 is configured to query the storage address and the first key of the image file corresponding to the task identifier from the task list stored in the local database.

The first sending module 300 is configured to send the first key to the storage address in the cloud, and the cloud is used to verify the first key, and after successful verification, returns the desensitized image file in the form of a binary stream.

The second receiving module 400 is configured to receive the desensitized image file returned by the cloud in the form of binary stream, and send the desensitized image file to the user terminal in the form of binary stream, so that the user terminal can parse the image file to obtain pixel information and desensitization information, and display the obtained pixel information and desensitization information.

The second sending module 500 is configured to receive structured data marked with pixel information by the user terminal, and upload the structured data to the cloud for associated storage with the image files stored in the cloud.

In one embodiment, the image data labeling apparatus further includes:

The third receiving module is configured to receive the labeling request sent by the user terminal, and query the task list stored in the first database according to the labeling request to obtain tasks whose status is the status to be labelled.

The third sending module is configured to send the task whose status is to be marked to the user terminal, and the task whose status is to be marked is used for selection by the user terminal.

In one embodiment, the image data labeling apparatus further includes:

The fourth receiving module is used for receiving the second key corresponding to the structured data returned by the cloud, storing the second key in the corresponding task in the task list, and modifying the state of the corresponding task in the task list to Pending review status.

The second query module is configured to receive the audit request sent by the audit terminal, and query the tasks in the task list that are in the state to be audited according to the audit request.

The selection module is used to obtain the tasks whose status is pending review and return them to the review terminal, and receive the selected tasks returned by the review terminal.

The audit module is used to query the second key corresponding to the selected task, and return the second key to the audit terminal. The second key is used for the audit terminal to download structured data from the cloud and conduct audit, and will audit The results are stored in the local database.

In one embodiment, the image data labeling apparatus further includes:

The storage module is used to send the task corresponding to the structured data whose audit result is passed and the corresponding second key to the R&D platform for storage, and the task and the second key stored in the R&D platform are used for the R&D terminal to query the task The corresponding second key, and the structured data corresponding to the audited task is downloaded from the cloud through the second key.

In one embodiment, the above-mentioned first query module 200 includes:

The load balancing unit is configured to select a processor for execution through a load balancing algorithm, and query the task address and the first key corresponding to the task identifier from the task list stored in the local database.

For the specific limitation of the image data labeling apparatus, please refer to the limitation on the image data labeling method above, which will not be repeated here. All or part of the modules in the above image data labeling apparatus can be implemented by software, hardware and combinations thereof. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

In one embodiment, a computer device is provided, and the computer device may be a server, and its internal structure diagram may be as shown in FIG. 5 . The computer device includes a processor, memory, a network interface, and a database connected by a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium, an internal memory. The nonvolatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the execution of the operating system and computer programs in the non-volatile storage medium. The database of the computer device is used to store image data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for labeling image data is implemented.

Those skilled in the art can understand that the structure shown in FIG. 5 is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer equipment to which the solution of the present application is applied. Include more or fewer components than shown in the figures, or combine certain components, or have a different arrangement of components.

In one embodiment, a computer device is provided, including a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the processor implements the following steps: receiving a task annotation request sent by a user terminal, and the task annotation request carries There is a task identification; from the task list stored in the local database, query the storage address and first key of the image file corresponding to the task identification; send the first key to the storage address in the cloud, and the cloud is used to perform the first key. Verification, and after successful verification, returns the desensitized image files in the form of binary streams; the receiving cloud returns the desensitized image files in the form of binary streams, and sends the desensitized image files in the form of binary streams to User terminal, so that the user terminal can analyze the image file to obtain pixel information and desensitization information, and display the obtained pixel information and desensitization information; receive the structured data marked by the user terminal on the pixel information, and upload the structured data to the cloud for storage in association with the image files stored in the cloud.

In one embodiment, before receiving the task labeling request sent by the user terminal when the processor executes the computer program, the method further includes: receiving the labeling request sent by the user terminal, and querying the task list stored in the first database according to the labeling request to obtain the status It is the task in the state to be marked; the task in the state to be marked is sent to the user terminal, and the task in the state to be marked is used for the user terminal to select.

In one embodiment, after the processor executes the computer program and uploads the structured data to the cloud for associated storage with the image files stored in the cloud, the method further includes: receiving a first image corresponding to the structured data returned by the cloud. Second key, store the second key in the corresponding task in the task list, and modify the status of the corresponding task in the task list to the pending review state; receive the review request sent by the review terminal, and query the task according to the review request The tasks whose status is pending review in the list; obtain the tasks whose status is pending review and return them to the review terminal, and receive the selected tasks returned by the review terminal; query the second key corresponding to the selected task, and assign the second key to the selected task. The key is returned to the audit terminal, and the second key is used for the audit terminal to download structured data from the cloud and conduct audit, and store the audit result in the local database.

In one embodiment, after the processor executes the computer program and stores the audit result in the local database, it further includes: sending the task corresponding to the structured data whose audit result is passed and the corresponding second key to the research and development platform The task and the second key stored in the R&D platform are used for the R&D terminal to query the second key corresponding to the task, and download the structured data corresponding to the approved task from the cloud through the second key.

In one embodiment, querying the task address and the first key corresponding to the task identifier from the task list stored in the local database when the processor executes the computer program includes: selecting the processor for execution through a load balancing algorithm, From the task list stored in the local database, query the task address and the first key corresponding to the task identifier.

In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented: receiving a task annotation request sent by a user terminal, and the task annotation request carries a task identifier ; From the task list stored in the local database, query the storage address and first key of the image file corresponding to the task identifier; send the first key to the storage address in the cloud, and the cloud is used to verify the first key, and After successful verification, the desensitized image file is returned in a binary stream; the receiving cloud returns the desensitized image file in a binary stream, and sends the desensitized image file to the user terminal in a binary stream. In order to make the user terminal analyze the image file to obtain the pixel information and desensitization information, and display the obtained pixel information and desensitization information; Associated storage with image files stored in the cloud.

In one embodiment, before receiving the task labeling request sent by the user terminal when the computer program is executed by the processor, the method further includes: receiving the labeling request sent by the user terminal, and querying the task list stored in the first database according to the labeling request to obtain The task whose status is the status to be marked; the task whose status is the status to be marked is sent to the user terminal, and the task whose status is the status to be marked is used for selection by the user terminal.

In one embodiment, after the computer program is executed by the processor, the structured data is uploaded to the cloud to be stored in association with the image files stored in the cloud, and further includes: receiving a data corresponding to the structured data returned by the cloud. The second key is stored in the corresponding task in the task list, and the state of the corresponding task in the task list is modified to the pending review state; the review request sent by the review terminal is received, and the query is made according to the review request In the task list, the state of the task is pending review; obtain the task whose status is pending review and return it to the review terminal, and receive the selected task returned by the review terminal; query the second key corresponding to the selected task, and send the second key to the selected task. The key is returned to the audit terminal, and the second key is used for the audit terminal to download structured data from the cloud for audit, and store the audit result in the local database.

In one embodiment, after the computer program is executed by the processor, after the audit result is stored in the local database, it further includes: sending the task corresponding to the structured data whose audit result is passed and the corresponding second key to the research and development The platform stores the tasks and the second key stored on the R&D platform for the R&D terminal to query the second key corresponding to the task, and download the structured data corresponding to the approved task from the cloud through the second key.

In one embodiment, when the computer program is executed by the processor, querying the task address and the first key corresponding to the task identifier from the task list stored in the local database includes: selecting the processor for execution through a load balancing algorithm , query the task address and the first key corresponding to the task identifier from the task list stored in the local database.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage In the medium, when the computer program is executed, it may include the processes of the above-mentioned method embodiments. Wherein, any reference to memory, storage, database or other medium used in the various embodiments provided in this application may include non-volatile and/or volatile memory. Nonvolatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Road (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

The technical features of the above embodiments can be combined arbitrarily. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction in the combination of these technical features It is considered to be the range described in this specification.

The above-mentioned embodiments only represent several embodiments of the present application, and the descriptions thereof are specific and detailed, but should not be construed as a limitation on the scope of the invention patent. It should be pointed out that for those skilled in the art, without departing from the concept of the present application, several modifications and improvements can be made, which all belong to the protection scope of the present application. Therefore, the scope of protection of the patent of the present application shall be subject to the appended claims.

Claims (10)

1. A method for labeling image data, the method comprising: receiving a task labeling request sent by the user terminal, where the task labeling request carries a task identifier; From the task list stored in the local database, query the storage address and first key of the image file corresponding to the task identifier; Sending the first key to the storage address in the cloud, where the cloud is used to verify the first key, and after successful verification, returns the desensitized image file in the form of a binary stream; Receive the desensitized image file returned by the cloud in a binary stream, and send the desensitized image file to the user terminal in a binary stream, so that the user terminal parses the image file to obtain pixel information and desensitization information, and display the obtained pixel information and desensitization information; Receive the structured data marked by the user terminal on the pixel information, and upload the structured data to the cloud for associated storage with the image files stored in the cloud. 2. The method according to claim 1, wherein before receiving the task labeling request sent by the user terminal, the method further comprises: receiving a labeling request sent by the user terminal, and querying the task list stored in the first database according to the labeling request to obtain tasks whose status is to be labelled; Sending the task whose status is the status to be marked to the user terminal, and the task whose status is the status to be marked is used for selection by the user terminal. 3 . The method according to claim 1 , wherein after uploading the structured data to the cloud and storing it in association with the image files stored in the cloud, the method further comprises: 3 . Receive the second key corresponding to the structured data returned by the cloud, store the second key in the task corresponding to the task list, and store the corresponding task in the task list. The status is changed to pending review; Receive the audit request sent by the audit terminal, and query the tasks in the task list whose status is to be audited according to the audit request; Obtaining the task whose status is to be reviewed and returning it to the review terminal, and receiving the selected task returned by the review terminal; Query the second key corresponding to the selected task, and return the second key to the audit terminal, where the second key is used for the audit terminal to download the structured data from the cloud And conduct audit, and store the audit result to the local database. 4. The method according to claim 3, wherein after storing the audit result in the local database, the method further comprises: Send the task and the corresponding second key corresponding to the structured data that the audit result is passed to the research and development platform for storage, and the task and the second key stored in the research and development platform are used for the research and development terminal to query the corresponding task. and download the structured data corresponding to the audited task from the cloud through the second key. 5. The method according to any one of claims 1 to 4, wherein, in the task list stored in the local database, querying the task address and the first key corresponding to the task identifier, comprising: The processor is selected for execution through a load balancing algorithm, and the task address and the first key corresponding to the task identifier are queried from the task list stored in the local database. 6. An image data labeling device, wherein the device comprises: a first receiving module, configured to receive a task labeling request sent by a user terminal, where the task labeling request carries a task identifier; a first query module, configured to query the storage address and the first key of the image file corresponding to the task identifier from the task list stored in the local database; The first sending module is used to send the first key to the storage address in the cloud, and the cloud is used to verify the first key, and after the verification is successful, it returns a binary stream to the address. post-sensitization image files; The second receiving module is configured to receive the desensitized image file returned by the cloud in the form of binary stream, and send the desensitized image file to the user terminal in the form of binary stream, so that the user terminal can parse the image file. The image file obtains pixel information and desensitization information, and displays the obtained pixel information and desensitization information; A second sending module, configured to receive structured data marked by the user terminal on the pixel information, and upload the structured data to the cloud to associate with the image files stored in the cloud storage. 7. The apparatus of claim 6, wherein the apparatus further comprises: A third receiving module, configured to receive a labeling request sent by the user terminal, and query the task list stored in the first database according to the labeling request to obtain tasks whose status is the status to be labelled; The third sending module is configured to send the task whose status is to be marked to the user terminal, and the task whose status is to be marked is used for selection by the user terminal. 8. The apparatus of claim 6, wherein the apparatus further comprises: a fourth receiving module, configured to receive the second key corresponding to the structured data returned by the cloud, store the second key in the corresponding task in the task list, and store the second key in the task list. The status of the corresponding task in the task list is changed to pending review; A second query module, configured to receive an audit request sent by the audit terminal, and query the tasks in the task list that are in a state to be audited according to the audit request; a selection module, configured to obtain the task whose status is to be reviewed and return it to the review terminal, and receive the selected task returned by the review terminal; an audit module, configured to query the second key corresponding to the selected task, and return the second key to the audit terminal, where the second key is used for the audit terminal to download from the cloud The structured data is audited, and the audit results are stored in the local database. 9. A computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein the processor implements the steps of the method according to any one of claims 1 to 5 when the processor executes the computer program . 10. A computer-readable storage medium on which a computer program is stored, characterized in that, when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images