Data processing method, device and related equipment for training artificial intelligence models

Abstract

The present disclosure provides a data processing method, apparatus, and related equipment for training artificial intelligence models. The present disclosure relates to the field of computer technology, and in particular to technical fields such as intelligent interaction and databases. A specific implementation scheme is as follows: based on a first target ETL pipeline that supports visual configuration, a set of data to be processed in a source database is processed to separate tasks to be labeled that are suitable for storage in a target database from the source database; data is stored in the source database in the form of files; data is stored in the target database in the form of data tables; based on the target database, the tasks to be labeled are labeled to obtain labeled data; based on a second target ETL pipeline, the labeled data is processed and stored in the source database.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to technical fields such as intelligent interaction and database. Background Art

[0002] Currently, training AI models requires a large amount of data, so distributed file systems are often considered for this data. In addition, the data required for AI model training often requires a lot of manual labeling. Summary of the Invention

[0003] The present disclosure provides a data processing method, apparatus, and related equipment for training artificial intelligence models to solve or alleviate one or more technical problems in the prior art.

[0004] In a first aspect, the present disclosure provides a data processing method for training an artificial intelligence model, comprising:

[0005] Processing a set of data to be processed in a source database based on a first target ETL pipeline supporting visual configuration to separate tasks to be labeled from the source database that are suitable for storage in a target database; the source database stores data in the form of files; and the target database stores data in the form of data tables;

[0006] Based on the target database, labeling the task to be labeled to obtain labeled data;

[0007] The labeled data is processed based on a second target ETL pipeline, and the labeled data is stored in the source database.

[0008] In a second aspect, the present disclosure provides a data processing device for training an artificial intelligence model, comprising:

[0009] a first processing module, configured to process a set of to-be-processed data in a source database based on a first target ETL pipeline supporting visualization configuration, so as to separate from the source database to-be-annotated tasks suitable for storage in a target database; the source database stores data in the form of files; and the target database stores data in the form of data tables;

[0010] A labeling module, configured to label the task to be labeled based on the target database to obtain labeled data;

[0011] The second processing module is configured to process the labeled data based on a second target ETL pipeline and store the labeled data in the source database.

[0012] According to a third aspect, an electronic device is provided, including:

[0013] at least one processor; and

[0014] a memory communicatively connected to the at least one processor; wherein,

[0015] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform any method in the embodiments of the present disclosure.

[0016] In a fourth aspect, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute any method according to the embodiments of the present disclosure.

[0017] In a fifth aspect, a computer program product is provided, comprising a computer program, which implements any method according to the embodiments of the present disclosure when executed by a processor.

[0018] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the multiple drawings represent the same or similar components or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings only depict some embodiments provided in accordance with the present disclosure and should not be regarded as limiting the scope of the present disclosure.

[0020] Figure 1 is a schematic diagram of a scenario of a data processing method for training an artificial intelligence model according to the first embodiment of the present disclosure;

[0021] Figure 2 is a flowchart of a data processing method for training an artificial intelligence model according to the second embodiment of the present disclosure;

[0022] Figure 3 is a flowchart of a data processing method for training an artificial intelligence model according to the third embodiment of the present disclosure;

[0023] Figure 4 is a schematic diagram of a visual interface of a data processing method for training an artificial intelligence model according to a fourth embodiment of the present disclosure;

[0024] Figure 5 is a schematic diagram of a visual interface of a data processing method for training an artificial intelligence model according to a fifth embodiment of the present disclosure;

[0025] Figure 6is a schematic diagram of a visual interface of a data processing method for training an artificial intelligence model according to a sixth embodiment of the present disclosure;

[0026] Figure 7 is a schematic diagram of a visual interface of a data processing method for training an artificial intelligence model according to the seventh embodiment of the present disclosure;

[0027] Figure 8 is a schematic diagram of a visual interface of a data processing method for training an artificial intelligence model according to an eighth embodiment of the present disclosure;

[0028] Figure 9 is a schematic diagram of a visual interface of a data processing method for training an artificial intelligence model according to a ninth embodiment of the present disclosure;

[0029] Figure 10 is a flowchart of a data processing method for training an artificial intelligence model according to the tenth embodiment of the present disclosure;

[0030] Figure 11 is a schematic diagram of a visual interface of a data processing method for training an artificial intelligence model according to the eleventh embodiment of the present disclosure;

[0031] Figure 12 is a schematic diagram of a visual interface of a data processing method for training an artificial intelligence model according to the twelfth embodiment of the present disclosure;

[0032] Figure 13 is a schematic diagram of a data processing method for training an artificial intelligence model according to the thirteenth embodiment of the present disclosure;

[0033] Figure 14 is a structural diagram of a data processing device for training an artificial intelligence model according to the fourteenth embodiment of the present disclosure;

[0034] Figure 15 This is a block diagram of an electronic device used to implement the data processing method / home model design method for training an artificial intelligence model according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0035] The present disclosure will be described in further detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise indicated.

[0036] In addition, numerous specific details are provided in the following detailed description to better illustrate the present disclosure. Those skilled in the art will appreciate that the present disclosure can be practiced without certain specific details. In some instances, methods, means, components, circuits, etc. well known to those skilled in the art are not described in detail in order to highlight the main purpose of the present disclosure.

[0037] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of such features. Throughout the present disclosure, "plurality" means two or more, unless otherwise specifically defined.

[0038] It should be noted that, unless it is explicitly stated that there is a sequence of execution between different operations shown in the flowchart in the embodiments of the present disclosure, or there is a sequence of execution between different operations in technical implementation, otherwise, the execution order between multiple operations may not be prioritized, and multiple operations may also be executed simultaneously.

[0039] Since distributed file systems can store large amounts of data, unlabeled data used to train artificial intelligence models are generally stored using distributed file systems.

[0040] In order to quickly and conveniently label unlabeled data in a distributed file system, the present disclosure involves data transmission between two databases, including a source database and a target database. The source database stores data in the form of files, an example of which may be HDFS (Hadoop Distributed File System). HDFS is a highly fault-tolerant distributed file system for large-scale data storage. As a core component of the Hadoop ecosystem, HDFS uses a master-slave structure to divide data into multiple blocks and stores them in a distributed manner on multiple nodes, supporting data redundancy and fault tolerance across servers. HDFS is suitable for storing large-scale batch data, but there is a problem of updating part of the data and requiring a full update.

[0041] The target database stores data in the form of tables. An example target database might be MySQL, an open-source relational database management system that supports the storage and management of structured data. It uses SQL (Structured Query Language) for data addition, deletion, modification, and query operations, and is widely used in data persistence and transaction processing scenarios in the internet, finance, e-commerce, and other fields. Compared to databases that store data in file format, MySQL databases are easier to update data, but they may have a smaller storage capacity.

[0042] In the embodiment of the present disclosure, a data processing method for training an artificial intelligence model is proposed. Figure 1 The figure shows a schematic diagram of the scenario in which this method is applied. Figure 1 It includes a server 11 and a terminal device 12.

[0043] The terminal device 12 and the server 11 are connected via a wireless or wired network. The terminal device 12 includes, but is not limited to, electronic devices such as desktop computers, mobile phones, mobile computers, tablet computers, media players, smart wearable devices, and smart televisions. The server 11 can be a single server, a server cluster consisting of multiple servers, or a cloud computing center. The server 11 can be a standalone physical server, a server cluster consisting of multiple physical servers, or a distributed system. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms.

[0044] In the embodiment of the present disclosure, the terminal device 12 is supported to configure the ETL (Extract, Transform, Load) pipeline and provide it to the source database and the target database in the server 11 for data transmission. The terminal device 12 is also used to provide a visual interface and perform annotation processing on the data transmitted to the target database.

[0045] like Figure 2 FIG. 1 is a flow chart of a data processing method for training an artificial intelligence model provided by an embodiment of the present disclosure, including the following contents:

[0046] S201, based on the first target ETL pipeline supporting visualization configuration, the set of data to be processed in the source database is processed to separate the tasks to be labeled suitable for storage in the target database from the source database; the source database stores data in the form of files; the target database stores data in the form of data tables.

[0047] During implementation, the data set to be processed originally stored in the source database needs to be transferred to the target database. Therefore, an ETL pipeline is established between the source database and the target database for data transmission between the two databases.

[0048] ETL pipelines configured through a visual interface allow non-technical personnel to configure parameters and complete data processing workflows directly within the interface, reducing reliance on professional developers and lowering the barrier to entry. Furthermore, visual configuration allows users to quickly configure and adjust ETL pipelines. Modifications and adjustments within the visual interface make the entire configuration process more intuitive and efficient.

[0049] S202: Based on the target database, label the task to be labeled to obtain labeled data.

[0050] Labeling tasks involves manually or automatically adding auxiliary information to raw data (such as images, text, speech, and video) for training machine learning models. This information includes, but is not limited to, category labels, bounding boxes, key points, semantic segmentation labels, sentiment, and named entities. Data labeling is a prerequisite for building supervised learning models, and its quality has a critical impact on model performance. It is widely used in fields such as computer vision, natural language processing, and speech recognition.

[0051] In the home furnishing field, the types of labeling tasks include but are not limited to marking the position of the home model in the image (for example, using a bounding box to mark the position of the sofa in the image), marking the category of home product images (such as living room furniture, bedroom furniture, kitchen utensils, etc.), and the attributes of the home model (for example, marking the color, material, style, etc. of the sofa).

[0052] S203: Process the labeled data based on the second target ETL pipeline and store the labeled data in the source database.

[0053] The second target ETL pipeline performs the inverse of the first target ETL pipeline. For example, the first target ETL pipeline is responsible for transferring data from the source database to the target database, while the second target ETL pipeline is responsible for transferring data from the target database to the source database and storing it according to the requirements of the corresponding fields in the source database, including but not limited to naming and storage methods.

[0054] In the disclosed embodiments, since the source database is stored in file format, updating any field in this type of database requires updating the entire file. Therefore, this type of database is incompatible with and cannot support field-level annotation tasks. Alternatively, while this type of database supports field-level annotation tasks, updating it would waste significant computing resources. Furthermore, the source database typically stores a large amount of raw data, and frequent data updates performed directly on the source database can significantly impact its performance, slowing down query and write operations. Transferring data to the target database for annotation and updates via an ETL pipeline can reduce direct write operations to the source database. Furthermore, the ETL pipeline splits the data in the source database and stores it in the target database for annotation. The target database is stored in data tables, but this type of database has a smaller storage capacity and supports field-level updates. Therefore, processing and updating can be performed only on the task to be annotated, rather than performing large-scale updates on the entire database. This significantly reduces the amount of data that needs to be updated, thereby reducing the system resource usage during the update process. Furthermore, the processing flow of the first target ETL pipeline can be well adapted to the storage structure characteristics of different databases. Through the first target ETL pipeline, data in file form can be converted into a structure suitable for data table storage, or the data in the data table can be converted back to file form through the second target ETL pipeline, thereby achieving seamless conversion between the two storage structures, giving full play to their respective advantages, and improving the flexibility of data storage and management. In addition, storing the tasks to be labeled in the data table of the target database provides a structured and clear labeling environment for labelers. The form of the data table is convenient for labelers to understand and operate, can improve the efficiency and accuracy of labeling, and is also conducive to the management of the labeling process. In summary, the process based on the ETL pipeline strengthens the collaborative relationship between the source database and the target database, breaks the data silos, and realizes the flow and sharing of data between different systems. At the same time, the solution proposed in the embodiment of the present disclosure helps to build a more integrated and collaborative data ecosystem, and improves the data processing capabilities and business value of the entire system.

[0055] In some embodiments, the steps of visually configuring the first target ETL pipeline are roughly divided into four steps: configuring basic information, configuring data destination, configuring annotation types, and configuring personnel management. These four steps are described in detail below.

[0056] In some embodiments, the first target ETL pipeline supporting visualization configuration processes the data set to be processed in the source database to separate the tasks to be labeled suitable for storage in the target database from the source database, such as Figure 3 As shown, it can be implemented as follows:

[0057] S301 : In response to a first configuration operation on a to-be-processed data set in a source database in a first visualization interface, obtain at least one source field in the to-be-processed data set.

[0058] An exemplary first visual interface is as follows: Figure 4 As shown in the figure, it is the basic configuration information in the above steps. The task name can be set based on the actual situation. Figure 4 As shown, it can be set as a test task. Figure 4 The task description in can be used to configure the relevant content of the task. For example, it can be used to configure what content needs to be marked in the task, what type of marked task belongs to, etc. In addition, if the task description does not need to be filled in, it can also be left blank. The data source represents the name of the data set to be processed in the source database. In addition, the location information of the data set to be processed can also be uploaded as the data source. After uploading the data set to be processed, at least one source field can be parsed from the data set to be processed. Figure 4 As shown in , the first row is at least one source field, which exemplarily includes designId (used to indicate the name of the design), k2d_vis (used to indicate the display method, coordinates, style and other details of the data to be annotated in two-dimensional space), etc.

[0059] S302, in response to a second configuration operation on a specified source field in the at least one source field in the second visualization interface, determining the field attributes of at least one target field corresponding to the specified source field in the target database, and obtaining tasks to be labeled for each target field.

[0060] During implementation, the second visualization interface includes the aforementioned configuration data destination and the operation interface for configuring the annotation type. Users can configure the corresponding attributes of any source field in the target database in the second visualization interface, and then generate target fields and corresponding tasks to be annotated based on these attributes.

[0061] In the embodiment of the present disclosure, the method of obtaining source fields through a visual interface can more flexibly and conveniently obtain the source fields in the data set to be processed and determine the target field attributes in the target database, compared to the traditional method of using coding to obtain source fields, and provide a good foundation for the subsequent splitting of labeling tasks according to different needs. In addition, when business needs change, the use of coding requires professionals to rewrite and debug the code, which is time-consuming and labor-intensive. The visual configuration method can modify the corresponding configuration in the interface to quickly adjust the data processing flow and field mapping rules. Therefore, obtaining the task to be labeled based on the visual interface can lower the technical threshold and improve flexibility and adaptability.

[0062] In some embodiments, in response to a second configuration operation on a specified source field in the at least one source field in the second visualization interface, determining the field attributes of at least one target field corresponding to the specified source field in the target database, and obtaining tasks to be labeled for each target field can be specifically implemented as follows:

[0063] Step A1: In response to the first sub-operation of creating a target table name in the second visual interface, a target data table corresponding to the target table name is created in the target database.

[0064] The first sub-operation in the second visual interface is to create a target table name. The target table name refers to the name of the table created in the target database for storing specific data. It is a unique identifier for the target data table and is used to distinguish different data tables to facilitate data management and operation. The target data table is a structured table that actually stores data in the target database. It consists of multiple fields (columns), and each field has a specific data type and attribute. The target data table is used to store the data set to be processed extracted from the source database, providing a data basis for subsequent labeling tasks. Figure 5 As shown in the figure, this is the operation of configuring the data destination in the previous steps. Figure 5 The target table location in the table can be used to enter the location information for storing the data set to be processed. The data source is similar to the above, which is equivalent to the location information of the data to be processed in the source database.

[0065] Step A2, in response to the second sub-operation of configuring the field attributes of the target field for the specified source field in the second visual interface, mapping the field attributes of the specified source field to the target data table to obtain the tasks to be labeled for each target field;

[0066] The field attributes of the target field include: target field index, target field name, target field type, target field value, annotation type, and display type;

[0067] The display type includes any of the following: picture type, web page type, and 3D home decoration model type.

[0068] The image type refers to the task being annotated, presented to the user as images (e.g., a list of thumbnails). During the annotation task, users can refer to the images and annotate them. This includes operations such as drawing bounding boxes and marking key points. Users can also annotate the style and model of the furniture in the image.

[0069] The web page type refers to the task being annotated being presented to the user as a web page. This type is typically used to display information such as text and tables. In annotation tasks, the web page type can be used to correctly display the data to be annotated.

[0070] The 3D home model type presents the user with a 3D model of the object being annotated. This type is typically used to showcase 3D spatial information such as interior and architectural designs. This type accurately and intuitively displays the 3D structure and spatial relationships of a home model. This annotation task allows you to annotate furniture and decorative items within a 3D home model, including their location, size, and material.

[0071] The target field attributes include the target field index, target field name, and target field type. Figure 5 Click the "Add target field" control in the dialog box, and a pop-up window will pop up. Figure 6 You can define the target field index, target field name, and target field type. The target field index is a number used to uniquely identify the target field in the target database. The target field name is the name of the target field in the target database, used to identify and distinguish it from other fields. The target field type refers to the type of data that can be stored in the target field. If the data to be processed is text, the target field type can be TEXT.

[0072] In addition, the fields in the source database may contain fields that are irrelevant to the task to be labeled or have poor data quality. The task to be labeled usually only needs to focus on specific parts of it, rather than all fields. Different types of tasks to be labeled have different data requirements. According to the specific task requirements, only mapping the necessary fields can customize the target data table to make it more suitable for the task requirements and reduce the storage of excessive redundant data. Therefore, only a specified part of the source field names in the source database can be mapped. Of course, if the task requirements require all source field names to be mapped, all of its source field names can be mapped to target fields.

[0073] It should be noted that the primary key needs to be defined at the target field index. If any target field index is not defined as the primary key, any configured target field will be selected as the primary key to ensure the integrity of the source data.

[0074] The target field's field properties include the target field value, annotation type, and display type, which can be based on Figure 7 Define it, where the target field value is any of the previously configured target field values. The display type is the format in which the source data is displayed, which can be determined based on the actual situation. The annotation type is the type of annotation task to be annotated. Some annotation types can be pre-created, and here you can select from the names of the annotation types corresponding to pre-created annotation tasks.

[0075] Can be based on Figure 8In the visual interface, you can create annotation tasks. You need to define the name of the annotation type, the type of annotation operation, and the annotation operation content. The annotation type name represents the name of the annotation task and is used to distinguish it from other annotation tasks. The annotation operation type indicates whether the annotation operation is a normal annotation (i.e., text annotation) or a partitioned image annotation (dividing the image into multiple regions and annotating each region separately). For the annotation operation content, you need to specify whether it is required, the annotation item identifier, the annotation item name, the annotation item type, the annotation component type, the default value, and the input value range and format. The "Required" annotation item indicates whether the annotation item is required. This is typically indicated by "Yes" or "No." The "Annotation Item Identifier" is a unique identifier for the annotation item, used for internal identification and management. It is usually a short string such as 1, 2, or 3, but can also be key1, key2, or key3. The "Annotation Item Name" is a human-readable name for the annotation item, displayed to annotators in the annotation interface. It should be concise and clearly describe the content of the annotation item. "Annotation item type" is used to indicate the data type of the annotation item, such as text, number, enumeration, etc. "Annotation component type" is used to indicate the input component type used by the annotation item in the annotation interface, such as text box, drop-down menu, check box, etc. "Default value" is used to indicate the default input of the annotation item when there is no manual input. It can improve annotation efficiency and reduce the workload of the annotator. "Limit input value range and format" is used to indicate the restriction of the input value of the annotation item and specify the allowed value range or format. The specific setting is based on the actual situation. For example, it can be the annotation of the style of interior furniture and interior dimensions.

[0076] Each annotation task can be based on Figure 8 Create to get the name of the annotation task and annotation type, each time based on Figure 8 A labeling task can be created. When N labeling tasks are created, Figure 7 In the process, the name of the annotation type corresponding to the required target annotation task can be selected from N annotation tasks as the annotation type of the target field, where N is a positive integer greater than 0.

[0077] In addition, you can also set the preview thumbnail column name and preview title name based on actual conditions. Since you usually need to annotate multiple tasks, during the annotation process, you can display unannotated tasks as thumbnails or lists. In this case, you need to specify the preview thumbnail column name, that is, determine which target field to display, and the preview title name used to display the field.

[0078] In the embodiment of the present disclosure, by mapping the field attributes of the specified source field to the target data table, the data in the source database can be stored in the target database. At the same time, the field attributes of the target field are configured to obtain clear and specific tasks to be labeled, which lays the foundation for the subsequent generation of labeling tasks and improves the accuracy of generating tasks to be labeled.

[0079] In some embodiments, many users are unfamiliar with which display types are suitable for different data types. Leaving them to figure this out on their own would waste considerable time and effort. Therefore, to ensure that each piece of data to be processed is displayed in an appropriate manner, the following steps can be implemented: analyzing the target field value corresponding to the target field to obtain a recommended list of display types that are suitable for the target field value; and displaying the recommended list corresponding to the target field in the second visualization interface.

[0080] During implementation, if the source data is in a format such as jpg or png, it can be determined that the data to be processed is likely an image, and therefore the recommended display type list may be image. If the source data is http.www.com, it can be determined that the data to be processed is likely a webpage, and the recommended display type list may be webpage. Other types can be determined based on actual circumstances and are not limited in this embodiment.

[0081] In the disclosed embodiments, this approach eliminates the need to waste time selecting a display type; instead, the most appropriate display type can be directly selected from a recommended list. This allows for better adaptation to the characteristics of the data being processed and the annotation requirements, thereby reducing display errors caused by improperly selected display types. The resulting tasks to be annotated can then be displayed in a reasonable manner, improving annotation accuracy and efficiency.

[0082] In some embodiments, the annotation task scope of the annotation personnel may also be determined in response to a third configuration operation on the annotation personnel in the third visualization interface.

[0083] During implementation, the third visual interface is exemplarily as follows Figure 9 As shown in , it includes the annotator and the filtering conditions. The annotator is the person who performs the task to be annotated, and the filtering conditions are which tasks in the task to be annotated can be seen by the annotator. Figure 9 As shown, the target field value of the task to be labeled (i.e., the designld set above) can be used for division, GT means greater than, Figure 9 The annotation range in indicates that the tasks to be annotated greater than the 1000th task in the target field value are visible to the annotators.

[0084] In addition, you can also add conditions or condition groups based on actual conditions. For example, based on the "Add Condition" control, add another condition, and add designld less than 2000 to the condition group. In the case of AND in the group, it can be determined that the 1000th to 2000th tasks to be labeled in the target field value can be seen by the person to be labeled. In addition, when the relationship between the two conditions in the group is OR, any one of the two conditions that is met can be seen by the annotator. For example, when designld less than 2000 and designld greater than 1000 are created, all tasks to be labeled within less than 2000 can be seen by the annotator.

[0085] In addition, another conditional group can be created based on the "Add Conditional Group" control, which can be used to restrict another target field name, for example.

[0086] In the disclosed embodiment, the scope of labeling tasks is reasonably allocated according to the ability, experience or workload of the labelers, and the scope of labeling tasks of the labelers is reasonably determined, thereby realizing refined management of the labeling work, so that each labeler clearly knows the labeling content he is responsible for, and reducing problems such as duplication, omission or confusion of labeling tasks.

[0087] In some embodiments, when the task to be labeled is obtained from the target database, a visual representation json is configured to visualize the task to be labeled on the terminal device, so that subsequent labeling personnel can label the task to be labeled.

[0088] In some embodiments, based on the target database, the task to be labeled is labeled to obtain labeled data, such as Figure 10 Shown, including:

[0089] S1001, when the display type of the task to be labeled is a three-dimensional home decoration model type, the task to be labeled is parsed from the target database to obtain the three-dimensional home decoration model in the task to be labeled, each construction unit in the three-dimensional home decoration model, and the labeling status of each construction unit.

[0090] The 3D home model represents the entire house structure, where a building unit can be, for example, a bedroom, living room, kitchen, or any other space. It can also represent any component within the 3D home model, thus supporting component-level annotation. For example, kitchen cabinet doors, handles, and decorative lights.

[0091] S1002, output the annotation status of each construction unit, the three-dimensional home decoration model and the annotation task parameters of the three-dimensional home decoration model to the annotation interface; wherein the annotation status is marked or unmarked; wherein the annotation status of each construction unit is used to be displayed in the first display area of ​​the annotation interface; the three-dimensional home decoration model is used to be displayed in the second display area of ​​the annotation interface; the annotation task parameters are used to be displayed in the third display area of ​​the annotation interface; the second display area supports preset operations for each construction unit.

[0092] When implementing, if Figure 11 As shown, from left to right are the first display area, the second display area and the third display area.

[0093] The first display area displays images based on the target field value preset in the preview thumbnail column name and the title preset in the preview title name. If the display type corresponding to the target field value in the preview thumbnail column name is image, and the preview title name is preset as a picture list, the image list will be displayed in this area.

[0094] The second display area showcases the 3D home model. Annotators can perform preset operations, such as rotation, scaling, and translation, allowing them to observe and annotate the model from different angles. Furthermore, by clicking on any space in the 3D home model (such as the living room), the second display area displays that space and allows for preset operations, such as rotation, scaling, and translation.

[0095] The second display area also includes display settings, build management, build information, etc. Display settings refer to a series of adjustable options provided during the display of 3D home improvement models, which are used to control and optimize the visualization of the model so that the annotators can perform the annotation work more efficiently and accurately. Figure 11 As shown, these functions include hiding unit models, measuring angles, measuring areas, and measuring side lengths. The Hide Unit Model function makes part or all of the unit model invisible, leaving only the specific areas or building units that need to be annotated. The Measure Angle function measures and displays the angles between specified edges or faces in the model. The Measure Area function measures and displays the area of ​​specified areas or faces in the model. The Measure Side Length function measures and displays the length of specified sides in the model. Additionally, you can set measurement areas, lengths, and volumes based on your specific needs.

[0096] Building Management is used to manage the various building units in a 3D home model. For example, if you check the box for a floor, it will be visible; if you uncheck it, it will be invisible. In addition to floors, you can also set the visibility of walls, windows, doors, furniture, and more.

[0097] Building information displays detailed information about each building, such as the building's name, type (wall, floor, furniture, etc.), size, material, color, and other attributes. It can also include the intersection of two buildings, such as the intersection of two walls.

[0098] The third display area is for displaying the aforementioned preset annotation types, and the annotators are required to annotate in this area.

[0099] S1003 , in response to the marking operation on any construction unit, obtaining a marking result of any construction unit.

[0100] During implementation, the annotation personnel fill in the third display area based on various rules. For example, the size range can be in the form of a digital input box or a slider.

[0101] S1004: Based on the labeling result of any construction unit, update the tasks to be labeled in the target database.

[0102] During implementation, when the annotator completes the annotation of the target structure in the third display area, clicking the annotation completion control can classify it into the annotated part in the first display area.

[0103] S1005 , when it is determined that each construction unit of the three-dimensional home decoration model has been labeled, obtain labeled data of the three-dimensional home decoration model.

[0104] During implementation, when it is determined that all target construction and labeling are completed in the first display area (that is, the unlabeled tasks are 0), the labeled data of the three-dimensional home decoration model can be obtained, and then the labeled data of the three-dimensional home decoration model can be stored in the source database.

[0105] In the disclosed embodiment, after parsing the task to be annotated, the annotation status of each construction unit, the three-dimensional home improvement model, and the annotation task parameters are accurately pushed to the annotation interface. The annotator does not need to filter the required content from a large amount of information, and can directly work on the unannotated construction units, saving search time. The second display area supports preset operations on each construction unit, such as rotation, scaling, translation, etc., which makes it convenient for the annotator to observe and annotate the model from different angles, reduces the time wasted due to inconvenient operation, and further improves the annotation efficiency. In addition, the annotation interface is divided into different display areas, which respectively display the annotation status, three-dimensional home improvement model, and annotation task parameters. The entire layout is clear and intuitive, which makes it easy for the annotator to quickly obtain the required information, reduces operational errors and unnecessary operation steps caused by interface confusion, and improves the user experience.

[0106] In some embodiments, processing the labeled data based on the second target ETL pipeline and storing the labeled data in the source database includes:

[0107] Step B1: In response to the third sub-operation of creating a target partition in the fourth visualization interface, a target partition for storing labeled data is created in the source database.

[0108] When implementing, if Figure 12 As shown in the figure, enter the directory information of the target partition in the target partition to complete the partition creation operation.

[0109] Step B2, in response to the fourth sub-operation of configuring the field attributes of the source field for the target field in the fourth visualization interface, the field attributes of the target field are mapped to the target partition to obtain at least one labeled data suitable for storage in the source database.

[0110] Since there may be differences in the field names used between the target database and the source database, the fields need to be converted so that they can be stored in the source database.

[0111] Step B3: merge at least one piece of labeled data and store it in the target partition of the source database.

[0112] Exemplarily, the labeling task of each target construction in the aforementioned three-dimensional home improvement model is a separate piece of labeled data, and all the labeled data corresponding to the three-dimensional home improvement model can be stored in a target partition.

[0113] In the disclosed embodiment, by creating target partitions in the source database and mapping field attributes, refined organization and management of labeled data is achieved. Each labeled data has a clear storage location and corresponding relationship, which facilitates subsequent data query, statistics, and analysis.

[0114] In the embodiments of the present disclosure, a schematic diagram of a data processing method for training an artificial intelligence model is provided, such as Figure 13 As shown, it includes a source database 1301, a target database 1302, a terminal device 1303, a first target ETL pipeline 1304, and a second target ETL pipeline 1305. Among them, the source database 1301 processes the data to be processed into tasks to be labeled through the first target ETL pipeline 1304 and sends them to the data table in the target database 1302 for storage. The target database 1302 visualizes the tasks to be labeled on the terminal device 1303 based on json technology. After the labeling personnel complete the labeling of the tasks to be labeled, they obtain the labeled data. After the labeled data is stored in the target database 1302, it is stored in the source database 1301 through the second target ETL pipeline 1305. Among them, Figure 13The solid line portion shown in the figure is the flow process of the data to be processed in the source database, and the dotted line portion is the flow process of storing the marked data back to the source database.

[0115] Based on the same technical concept, the embodiment of the present disclosure also proposes a data processing device 1400 for training an artificial intelligence model, such as Figure 14 Shown, including:

[0116] The first processing module 1401 is configured to process the data set to be processed in the source database based on the first target ETL pipeline supporting visualization configuration, so as to separate the unlabeled tasks suitable for storage in the target database from the source database; the data is stored in the source database in the form of files; and the data is stored in the target database in the form of data tables;

[0117] A labeling module 1402 is used to label the task to be labeled based on the target database to obtain labeled data;

[0118] The second processing module 1403 is configured to process the labeled data based on the second target ETL pipeline and store the labeled data in the source database.

[0119] In some embodiments, the first processing module includes:

[0120] A first configuration unit is configured to obtain at least one source field in the data set to be processed in response to a first configuration operation on the data set to be processed in the source database in the first visualization interface;

[0121] The second configuration unit is used to respond to the second configuration operation of the specified source field in the at least one source field in the second visualization interface, determine the field attributes of at least one target field corresponding to the specified source field in the target database, and obtain the tasks to be labeled for each target field.

[0122] In some embodiments, the second configuration unit includes:

[0123] a creating subunit, configured to create a target data table corresponding to the target table name in the target database in response to a first sub-operation of creating the target table name in the second visual interface;

[0124] a mapping subunit, configured to, in response to a second sub-operation of configuring the field attributes of the target field for the specified source field in the second visual interface, map the field attributes of the specified source field to the target data table to obtain tasks to be labeled for each target field;

[0125] The field attributes of the target field include: target field index, target field name, target field type, target field value, annotation type, and display type;

[0126] The display type includes any of the following: picture type, web page type, and 3D home decoration model type.

[0127] In some embodiments, the system further includes an analysis module for:

[0128] Analyze the target field value corresponding to the target field and obtain a recommended list of display types that are suitable for the target field value;

[0129] A recommendation list corresponding to the target field is displayed in the second visual interface.

[0130] In some embodiments, a scope configuration module is further included to:

[0131] In response to a third configuration operation on the labeler in the third visualization interface, a labeling task scope of the labeler is determined.

[0132] In some embodiments, the annotation module includes:

[0133] a parsing unit, configured to parse the task to be labeled from the target database when the display type of the task to be labeled is a 3D home decoration model type, and obtain the 3D home decoration model in the task to be labeled, each building unit in the 3D home decoration model, and the labeling status of each building unit;

[0134] An output unit is used to output the annotation status of each building unit, the three-dimensional home decoration model, and the annotation task parameters of the three-dimensional home decoration model to the annotation interface; wherein the annotation status is marked or unmarked; wherein the annotation status of each building unit is used to be displayed in the first display area of ​​the annotation interface; the three-dimensional home decoration model is used to be displayed in the second display area of ​​the annotation interface; and the annotation task parameters are used to be displayed in the third display area of ​​the annotation interface; the second display area supports preset operations for each building unit;

[0135] a marking unit, configured to obtain a marking result of any building unit in response to a marking operation on any building unit;

[0136] An updating unit, used to update the tasks to be labeled in the target database based on the labeling results of any building unit;

[0137] The determination unit is used to obtain the labeled data of the three-dimensional home decoration model when it is determined that each construction unit of the three-dimensional home decoration model has been labeled.

[0138] In some embodiments, the second processing module includes:

[0139] a creating unit, configured to create a target partition for storing the labeled data in the source database in response to a third sub-operation of creating a target partition in the fourth visualization interface;

[0140] a mapping unit, configured to, in response to a fourth sub-operation of configuring the field attributes of the source field for the target field in the fourth visualization interface, map the field attributes of the target field to the target partition to obtain at least one piece of labeled data suitable for storage in the source database;

[0141] Merge at least one piece of labeled data and store it in the target partition of the source database.

[0142] For the description of specific functions and examples of each module and submodule of the device in the embodiment of the present disclosure, please refer to the relevant description of the corresponding steps in the above method embodiment, which will not be repeated here.

[0143] In the technical solutions disclosed herein, the acquisition, storage, and application of user personal information involved comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0144] Figure 15 FIG. 1 is a structural block diagram of an electronic device according to an embodiment of the present disclosure. Figure 15 As shown, the electronic device includes: a memory 1510 and a processor 1520. The memory 1510 stores a computer program that can be executed on the processor 1520. The number of memories 1510 and processors 1520 can be one or more. The memory 1510 can store one or more computer programs. When the one or more computer programs are executed by the electronic device, the electronic device performs the method provided by the above method embodiment. The electronic device may also include: a communication interface 1530 for communicating with external devices and performing data exchange.

[0145] If the memory 1510, the processor 1520, and the communication interface 1530 are implemented independently, the memory 1510, the processor 1520, and the communication interface 1530 can be connected to each other via a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 15 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0146] Optionally, in a specific implementation, if the memory 1510, the processor 1520 and the communication interface 1530 are integrated on a chip, the memory 1510, the processor 1520 and the communication interface 1530 can communicate with each other through an internal interface.

[0147] It should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. It is worth noting that the processor may be a processor that supports the Advanced RISC Machines (ARM) architecture.

[0148] Furthermore, optionally, the above-mentioned memory may include a read-only memory and a random access memory, and may also include a non-volatile random access memory. The memory may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may include a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may include a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available. For example, static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM) and direct memory bus random access memory (DR RAM).

[0149] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, 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 instructions are loaded and executed on a computer, the process or function described in the embodiment of the present disclosure 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 one website, computer, server or data center to another website, computer, server or data center via a wired (e.g., coaxial cable, optical fiber, data subscriber line (DSL)) or wireless (e.g., infrared, Bluetooth, 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 integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, or a magnetic tape), an optical medium (e.g., a digital versatile disc (DVD)), or a semiconductor medium (e.g., a solid-state drive (SSD)). It is worth noting that the computer-readable storage medium mentioned in the present disclosure may be a non-volatile storage medium, in other words, a non-transient storage medium.

[0150] Those skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware, or by a program to instruct the relevant hardware, and the program may be stored in a computer-readable storage medium, which may be a read-only memory, a disk, or an optical disk, etc.

[0151] In the description of the embodiments of the present disclosure, the reference terms "one embodiment," "some embodiments," "example," "specific example," or "some examples" mean that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present disclosure. Moreover, the specific features, structures, materials, or characteristics described may be combined in any appropriate manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless they are mutually inconsistent.

[0152] In the description of the embodiments of the present disclosure, unless otherwise specified, " / " means or. For example, A / B can mean A or B. "And / or" in this document is only a description of the association relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone.

[0153] In the description of the embodiments of the present disclosure, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of the present disclosure, unless otherwise specified, "plurality" means two or more.

[0154] The above description is merely an exemplary embodiment of the present disclosure and is not intended to limit the present disclosure. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present disclosure shall be included in the scope of protection of the present disclosure.

Claims

1. A data processing method for training an artificial intelligence model, comprising: Processing the to-be-processed data set in the source database based on the first target ETL pipeline supporting visualization configuration to separate to-be-annotated tasks suitable for storage in the target database from the source database; The source database stores data in the form of files; the target database stores data in the form of data tables; Based on the target database, labeling the task to be labeled to obtain labeled data; The labeled data is processed based on a second target ETL pipeline, and the labeled data is stored in the source database.

2. The method according to claim 1, wherein The first target ETL pipeline based on the visualization configuration supports processing the set of data to be processed in the source database to separate the tasks to be labeled suitable for storage in the target database from the source database, including: In response to a first configuration operation on the to-be-processed data set in the source database in a first visualization interface, obtaining at least one source field in the to-be-processed data set; In response to a second configuration operation on a specified source field in the at least one source field in the second visualization interface, field attributes of at least one target field corresponding to the specified source field in the target database are determined to obtain tasks to be labeled for each target field.

3. The method according to claim 2, wherein: In response to the second configuration operation on a designated source field in the at least one source field in the second visual interface, determining a field attribute of at least one target field corresponding to the designated source field in the target database, and obtaining tasks to be labeled for each target field, including: In response to a first sub-operation of creating a target table name in the second visual interface, creating a target data table corresponding to the target table name in the target database; In response to a second sub-operation of configuring the field attributes of the target field for the specified source field in the second visual interface, mapping the field attributes of the specified source field to the target data table to obtain tasks to be labeled for each target field; The field attributes of the target field include: target field index, target field name, target field type, target field value, annotation type, and display type; The display type includes any one of the following: picture type, web page type, and three-dimensional home decoration model type.

4. The method according to claim 3, further comprising: Analyze the target field value corresponding to the target field to obtain a recommended list of display types adapted to the target field value; The recommendation list corresponding to the target field is displayed in the second visual interface.

5. The method according to any one of claims 1 to 4, further comprising: In response to a third configuration operation on the labeler in the third visualization interface, a labeling task scope of the labeler is determined.

6. The method according to claim 1, wherein The step of labeling the task to be labeled based on the target database to obtain labeled data includes: When the display type of the task to be labeled is a three-dimensional home decoration model type, the task to be labeled is parsed from the target database to obtain the three-dimensional home decoration model in the task to be labeled, each building unit in the three-dimensional home decoration model, and the labeling status of each building unit; Outputting the annotation status of each building unit, the three-dimensional home improvement model, and the annotation task parameters of the three-dimensional home improvement model to the annotation interface; wherein the annotation status is annotated or unannotated; wherein the annotation status of each building unit is used to be displayed in a first display area of ​​the annotation interface; the three-dimensional home improvement model is used to be displayed in a second display area of ​​the annotation interface; the annotation task parameters are used to be displayed in a third display area of ​​the annotation interface; the second display area supports preset operations for each building unit; In response to a marking operation on any building unit, obtaining a marking result of the any building unit; Based on the labeling result of any one of the construction units, updating the task to be labeled in the target database; When it is determined that each construction unit of the three-dimensional home decoration model has been labeled, the labeled data of the three-dimensional home decoration model is obtained.

7. The method according to claim 3, wherein: The processing of the labeled data based on the second target ETL pipeline and storing the labeled data in the source database includes: In response to a third sub-operation of creating a target partition in a fourth visualization interface, creating a target partition in the source database for storing the labeled data; In response to a fourth sub-operation of configuring the field attributes of the source field for the target field in the fourth visualization interface, mapping the field attributes of the target field to the target partition to obtain at least one labeled data suitable for storage in the source database; The at least one piece of labeled data is merged and stored in a target partition of the source database.

8. A data processing device for training an artificial intelligence model, comprising: A first processing module is configured to process a set of data to be processed in a source database based on a first target ETL pipeline supporting visualization configuration, so as to separate tasks to be labeled suitable for storage in a target database from the source database; The source database stores data in the form of files; the target database stores data in the form of data tables; A labeling module, configured to label the task to be labeled based on the target database to obtain labeled data; The second processing module is configured to process the labeled data based on a second target ETL pipeline and store the labeled data in the source database.

9. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 7.

11. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 7.

Images