Interaction method and system of labeling software and training platform

Abstract

The invention provides an interaction method and system for labeling software and a training platform, and the method comprises the steps: obtaining a labeling project list from a plurality of labeling software through the training platform, and displaying the labeling project list on a human-computer interaction interface; the training platform performs data analysis on each item in the annotation project list to obtain annotation project information; the training platform selects any annotation project from the human-computer interaction interface; the training platform obtains corresponding labeling data from labeling software according to the selected labeling project; the training platform converts the annotation data into a data set; the training platform automatically calls the data set and the configuration data to train the AI model to obtain a training result; the training result comprises the trained model; the training platform automatically creates a reasoning process according to the trained model or the pre-trained model; the training platform creates an automatic labeling service for the labeling project selected by the user by calling a compatible interface method; and various labeling software is compatible, so that the development efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence and computer program application technology, and in particular to a method and system for interaction between annotation software and training platform. Background Technology

[0002] With the deep application of artificial intelligence technology in fields such as computer vision, natural language processing, speech recognition, and structured data analysis, the performance improvement of deep learning models increasingly depends on the support of high-quality, large-scale labeled datasets. Data labeling and model training have become core links in the artificial intelligence research and development process.

[0003] Currently, various professional annotation software programs have emerged in the industry for different data types (images, text, audio, video, etc.), such as LabelStudio and Diffgram, which can annotate multiple types of data.

[0004] In the actual model development process, there are still significant technical barriers and process fragmentation issues between the data annotation and model training stages of traditional technical solutions (non-integrated solutions), mainly reflected in the following three aspects: 1) Incompatible annotation data formats: Different annotation software uses fundamentally different annotation data storage formats, and their output file formats may only be one or a few, lacking a unified industry standard. When R&D teams need to use datasets generated by different annotation tools for training the same model, they must manually write format conversion scripts or use third-party tools for format adaptation. This not only increases additional labor costs but also easily leads to the loss of annotation information (such as coordinate offsets and label misalignment) due to format conversion errors, seriously affecting dataset quality and model training results. 2) Fragmented Data Transfer Process: In existing technical solutions, labeled datasets require manual downloading, storage migration (e.g., copying from the labeling software server to the training server), and manual import into training frameworks (e.g., TensorFlow, PyTorch) to achieve data transfer. The entire process relies on manual intervention and cannot form an automated "labeling-acquisition-training" closed loop. Especially in scenarios with large-scale datasets (e.g., million-image datasets), manual transfer is time-consuming (usually several hours to several days), significantly reducing model iteration efficiency and making it difficult to meet the needs of rapid research and development. 3) High difficulty in adapting training environments: In addition to format differences, datasets output by different annotation software may also have inconsistent data preprocessing standards (such as image size normalization, text encoding format, audio sampling rate, etc.). Existing model training platforms typically only support preprocessed data in specific formats. Researchers need to configure preprocessing procedures and training environment parameters separately for annotated data from different sources, resulting in poor universality of the training process and a high risk of training failure due to incorrect parameter configuration, thus increasing the technical threshold and R&D risks.

[0005] In summary, the aforementioned problems result in low collaborative efficiency between data annotation and model training in the current AI research and development process, high labor costs, long process cycles, and significant data quality risks. Summary of the Invention

[0006] In view of this, the purpose of this invention is to provide an interaction method and system between annotation software and training platform, which has low coupling between annotation and training, is compatible with multiple annotation software, and has strong scalability; it realizes an automated closed loop of "annotation-acquisition-training", reducing development time costs; and it realizes the function of directly starting automated annotation service, assisting users to perform data annotation faster and greatly improving development efficiency.

[0007] In a first aspect, embodiments of the present invention provide a method for interaction between annotation software and a training platform, the method comprising: The training platform obtains a list of labeled projects from multiple labeled software programs and displays the list of labeled projects on a human-computer interaction interface. The list of labeled projects includes multiple labeled projects. The training platform parses the data of each item in the list of labeled projects to obtain the labeled project information; The training platform selects any one of the labeled projects from the human-computer interaction interface; The training platform obtains the corresponding annotation data from the annotation software based on the selected annotation project; The training platform converts the labeled data into a dataset; The training platform will automatically call the dataset and configuration data to train the AI ​​model and obtain the training results; wherein, the training results include the trained model; The training platform automatically creates an inference process based on the trained model or pre-trained model. The training platform creates automated annotation services for the annotation project selected by the user by calling compatible interface methods.

[0008] Furthermore, the training platform obtains a list of labeled projects from multiple labeled software programs, including: The training platform sends a request for a list to the annotation software, so that the annotation software can obtain the list of annotation projects based on the request. The training platform receives the list of labeled projects sent by the labeling software.

[0009] Furthermore, the training platform obtains corresponding annotation data from the annotation software based on the selected annotation project, including: The training platform sends a request to the annotation software based on the selected annotation project, so that the annotation software can obtain the annotation data corresponding to the annotation project based on the request. The training platform receives the annotation data sent by the annotation software.

[0010] Furthermore, the training platform creates automated annotation services for the annotation project selected by the user by calling compatible interface methods, including: The training platform acquires the labeled project selected by the user on the human-computer interaction interface; The training platform configures automated annotation services in the annotation project corresponding to the annotation software by calling the compatible interface method.

[0011] Furthermore, the method also includes: Once the automated annotation service is successfully configured, the corresponding annotation project can be accessed through the human-computer interaction interface of the annotation software. The annotation software obtains the user's confirmation to start annotation; The annotation software begins automated annotation based on the confirmed annotation information and obtains the annotation results.

[0012] Secondly, embodiments of the present invention provide an interaction system between annotation software and a training platform, the system comprising: The annotation project list acquisition module is used by the training platform to acquire an annotation project list from multiple annotation software and display the annotation project list on the human-computer interaction interface. The annotation project list includes multiple annotation projects. The parsing module is used by the training platform to parse the data of each item in the list of labeled projects to obtain the labeled project information; The selection module is used by the training platform to select any one of the annotation projects from the human-computer interaction interface; The annotation data acquisition module is used by the training platform to acquire corresponding annotation data from the annotation software based on the selected annotation project; A conversion module is used by the training platform to convert the labeled data into a dataset; The training module is used by the training platform to automatically call the dataset and configuration data to train the AI ​​model and obtain training results; wherein, the training results include the trained model; An inference process creation module is used by the training platform to automatically create an inference process based on the trained model or pre-trained model. An automated annotation service creation module is used by the training platform to create automated annotation services for the annotation project selected by the user by calling compatible interface methods.

[0013] Furthermore, the module for obtaining the labeled project list is specifically used for: The training platform sends a request for a list to the annotation software, so that the annotation software can obtain the list of annotation projects based on the request. The training platform receives the list of labeled projects sent by the labeling software.

[0014] Furthermore, the labeled data acquisition module is specifically used for: The training platform sends a request to the annotation software based on the selected annotation project, so that the annotation software can obtain the annotation data corresponding to the annotation project based on the request. The training platform receives the annotation data sent by the annotation software.

[0015] Thirdly, embodiments of the present invention provide an electronic device, including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the computer program to implement the method described above.

[0016] Fourthly, embodiments of the present invention provide a computer-readable medium having processor-executable non-volatile program code that causes the processor to perform the method described above.

[0017] This invention provides a method and system for interaction between annotation software and a training platform, comprising: the training platform obtaining a list of annotation projects from multiple annotation software and displaying the list on a human-computer interaction interface, the list including multiple annotation projects; the training platform parsing each item in the list to obtain annotation project information; the training platform selecting any annotation project from the human-computer interaction interface; the training platform obtaining corresponding annotation data from the annotation software based on the selected annotation project; the training platform converting the annotation data into a dataset; the training platform automatically calling the dataset and configuration data to train the AI ​​model and obtain training results; wherein, the training results include the trained model; the training platform automatically creating an inference process based on the trained model or the pre-trained model; the training platform creating an automated annotation service for the annotation project selected by the user by calling a compatible interface method; the coupling between annotation and training is low, it is compatible with multiple annotation software, and has strong scalability; it realizes an automated closed loop of "annotation-acquisition-training", reducing development time costs; it realizes the function of directly starting the automated annotation service, assisting users to perform data annotation faster, and greatly improving development efficiency.

[0018] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained in accordance with the structures particularly pointed out in the description, claims and drawings.

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0020] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating the interaction method between the annotation software and the training platform provided in Embodiment 1 of the present invention. Figure 2 This is a schematic diagram illustrating the interaction and communication between the training platform and various annotation software provided in Embodiment 1 of the present invention. Figure 3 A schematic diagram showing the labeled project list for the training platform provided in Embodiment 1 of the present invention; Figure 4This is a schematic diagram illustrating how the training platform provided in Embodiment 1 of the present invention converts labeled projects into datasets. Figure 5 This is a schematic diagram of AI model training provided in Embodiment 1 of the present invention; Figure 6 This is a schematic diagram illustrating the creation and implementation of automated annotation as provided in Embodiment 1 of the present invention; Figure 7 This is a schematic diagram of the interaction system between the annotation software and the training platform provided in Embodiment 2 of the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] To facilitate understanding of this embodiment, the embodiments of the present invention will be described in detail below.

[0024] Example 1: Figure 1 This is a flowchart illustrating the interaction method between the annotation software and the training platform provided in Embodiment 1 of the present invention.

[0025] Reference Figure 1 The method includes the following steps: Step S101: The training platform obtains a list of annotation projects from multiple annotation software and displays the list of annotation projects on the human-computer interaction interface. The list of annotation projects includes multiple annotation projects. In step S102, the training platform parses the data of each item in the list of labeled projects to obtain the labeled project information; Specifically, refer to Figure 2 Once the training platform is started, it communicates with various annotation software through compatible interface methods such as API and HTTP, and obtains a list of annotation projects, thereby displaying all annotation projects to the user through a human-computer interaction interface.

[0026] The training platform parses the data of each item in the list of labeled projects to directly obtain the labeled project information, which includes the data type of the labeled project, such as object detection, image classification, image segmentation, speech recognition, etc.

[0027] Step S103: The training platform selects any annotation project from the human-computer interaction interface; Step S104: The training platform obtains the corresponding annotation data from the annotation software based on the selected annotation project; Step S105: The training platform converts the labeled data into a dataset; Specifically, the training platform's human-computer interaction interface provides relevant operation entry points, enabling the automatic conversion of annotation projects from the annotation platform into datasets for import into the training platform; the training platform obtains / downloads data from the annotation platform through compatible interface methods and saves it as a dataset that can be used by deep learning frameworks.

[0028] Reference Figure 4 Users select a labeling project on the training platform's user interface and then click "Convert to Dataset." The training platform requests data from the labeling software via a compatible interface method. The labeling software returns the labeled data, which the training platform then saves as a dataset that can be used by deep learning frameworks.

[0029] In step S106, the training platform will automatically call the dataset and configuration data to train the AI ​​model and obtain the training results; the training results include the trained model. Specifically, the training platform will automatically call the dataset and configuration data to train the AI ​​model; the configuration data includes pre-trained models or custom training hyperparameters.

[0030] At the training task entry point of the training platform, users can choose to train using different hardware types, including CPU, GPU, Ascend, etc. The human-computer interaction interface will display the current hardware usage based on the selection to help users better adjust training parameters.

[0031] Reference Figure 5 Users select a dataset and configure training parameters, hardware devices, and other information on the human-computer interaction interface of the training platform. The training platform will automatically call the dataset and configuration data of the deep learning framework for training and save the training results for users to view and use.

[0032] Step S107: The training platform automatically creates an inference process based on the trained model or pre-trained model. In step S108, the training platform creates an automated annotation service for the annotation project selected by the user by calling a compatible interface method.

[0033] Specifically, the training platform will store the trained models, results, and logs obtained from the training tasks, allowing users to view, download, and use them; the training platform provides a model download interface, thereby supporting automated deployment and model application software.

[0034] The training platform's user interface provides an entry point for creating automated annotation services, with each annotation project corresponding to a specific project. When a user selects a annotation project, they can access the automated annotation service creation portal. They can choose either a pre-trained model or a provided pre-trained model. The training platform will automatically create an inference process based on the selected model and call compatible interface methods to create an automated annotation service for the annotation project.

[0035] The training platform provides a list of automated annotation services for display and management, which can be started and stopped according to user needs, thereby allowing users to make greater use of hardware resources.

[0036] In response to the technical problems existing in the current technology, there is an urgent need for a technical solution that can be compatible with any annotation software, realize the automated acquisition and training of annotation data, and has the characteristics of low coupling, wide adaptability and high maturity, so as to break down the barriers between links and improve the overall R&D efficiency.

[0037] This application relates to the field of artificial intelligence and computer program application technology, specifically to the supporting technologies for deep learning model training. This technology focuses on solving the problem of the disconnect between data annotation and model training processes. By constructing standardized data interaction interfaces and adaptation layers, it achieves compatibility with various third-party annotation software (including but not limited to image annotation, text annotation, audio annotation, and other specialized tools), enabling automated acquisition of annotated datasets in different formats and completing the entire preprocessing and model training process. Its technical solution covers key technologies such as the architectural design of computer program products, big data format conversion and cleaning, and adaptation and optimization of deep learning training environments. It can be widely applied to model development scenarios in sub-fields of artificial intelligence such as computer vision, natural language processing, and speech recognition.

[0038] Furthermore, step S101 includes the following steps: Step S201: The training platform sends a request for a list to the annotation software so that the annotation software can obtain a list of annotation projects based on the request. Step S202: The training platform receives the list of labeled projects sent by the labeling software.

[0039] Specifically, refer to Figure 3 The training platform sends a request for a list to the annotation software through a compatible interface method. The annotation software returns a list of all annotation projects in the software, and the training platform's human-computer interaction interface displays these annotation projects to the user.

[0040] Furthermore, step S104 includes the following steps: Step S301: The training platform sends a request to the annotation software based on the selected annotation project, so that the annotation software can obtain the annotation data corresponding to the annotation project based on the request. Step S302: The training platform receives the annotation data sent by the annotation software.

[0041] Furthermore, step S108 includes the following steps: Step S401: The training platform obtains the annotation project selected by the user on the human-computer interaction interface; In step S402, the training platform configures automated annotation services in the annotation project corresponding to the annotation software by calling the compatible interface method.

[0042] Furthermore, the method also includes the following steps: Step S501: After successfully configuring the automated annotation service, enter the corresponding annotation project on the human-computer interaction interface of the annotation software. Step S502: The annotation software obtains the user's confirmation to start annotation; In step S503, the annotation software starts automatic annotation based on the confirmed annotation information and obtains the annotation results.

[0043] Specifically, refer to Figure 6 Users select the annotation project on the human-computer interaction interface of the training platform, and then enter the configuration window of the automated annotation backend. In this configuration window of the automated annotation backend, users select a model, which can be a pre-trained model or a model trained from a certain training result. The training platform creates an inference process based on the pre-trained model or the trained model.

[0044] The training platform carries information about the inference process and then requests the annotation software to configure automated annotation services. After the automated annotation services are successfully configured, the user enters the corresponding annotation project on the human-computer interaction interface of the annotation software and then confirms to start annotation. The annotation software will then begin automated annotation based on the configured automated annotation service information.

[0045] The effect achieved by this application is as follows: 1) The coupling between annotation and training is low, it is compatible with multiple annotation software, and has strong scalability; 2) It can adapt to a variety of data types / training tasks. It can connect to all current data types and training tasks as needed, and is directly compatible with the powerful ecosystem of current AI model training.

[0046] 3) It is compatible with multiple annotation software and has a high level of maturity, which improves the overall stability and makes it easy to handle large-scale data annotation and model training.

[0047] 4) It realizes an automated closed loop of "labeling-acquisition-training", reducing the development time cost.

[0048] 5) It enables the direct activation of automated annotation services, helping users to annotate data faster and greatly improving development efficiency.

[0049] 6) It provides a user-friendly human-computer interaction interface, allowing users to develop AI models without needing much professional knowledge.

[0050] 7) The training platform has implemented a one-to-many annotation software, and the annotation software can be deployed in a distributed manner with the training platform, which greatly improves development efficiency, achieves environment isolation, and ensures data security.

[0051] This application proposes an interaction method between annotation software and a training platform, enabling interactive communication between the training platform and various annotation software programs. The training platform can display annotation projects through a compatible interface method and can directly convert these projects into datasets for storage. This application can create inference processes based on pre-trained or trained models and create automatic annotation services through interactive communication with annotation software.

[0052] Example 2: Figure 7 This is an interaction system between annotation software and training platform provided in Embodiment 2 of the present invention.

[0053] Reference Figure 7 The system includes: The annotation project list acquisition module is used by the training platform to obtain an annotation project list from multiple annotation software and display the annotation project list on the human-computer interaction interface. The annotation project list includes multiple annotation projects. The parsing module is used by the training platform to parse the data of each item in the list of labeled projects to obtain the labeled project information; The selection module is used by the training platform to select any annotation project from the human-computer interaction interface; The annotation data acquisition module is used by the training platform to obtain the corresponding annotation data from the annotation software based on the selected annotation project; The conversion module is used by the training platform to convert labeled data into datasets; The training module is used to train the AI ​​model by automatically calling the dataset and configuration data of the training platform and obtaining the training results; the training results include the trained model. The inference process creation module is used by the training platform to automatically create inference processes based on the trained model or pre-trained model. The automated annotation service creation module is used by the training platform to create automated annotation services for the annotation project selected by the user by calling compatible interface methods.

[0054] Furthermore, the module for obtaining the project list is specifically used for: The training platform sends a request for a list to the annotation software so that the annotation software can obtain a list of annotation projects based on the request. The training platform receives a list of labeled projects sent by the annotation software.

[0055] Furthermore, the labeled data acquisition module is specifically used for: The training platform sends a request to the annotation software based on the selected annotation project, so that the annotation software can obtain the annotation data corresponding to the annotation project based on the request. The training platform receives annotation data sent by the annotation software.

[0056] This invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the interaction method between the annotation software and the training platform provided in the above embodiments.

[0057] This invention also provides a computer-readable medium having processor-executable non-volatile program code, on which a computer program is stored, and which, when run by a processor, executes the steps of the interaction method between the annotation software and the training platform described above.

[0058] The computer program product provided in this embodiment of the invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the preceding method embodiments. For specific implementation details, please refer to the method embodiments, which will not be repeated here.

[0059] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system and apparatus described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0060] Furthermore, in the description of the embodiments of the present invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention based on the specific circumstances.

[0061] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0062] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0063] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for interaction between annotation software and a training platform, characterized in that, The method includes: The training platform obtains a list of labeled projects from multiple labeled software programs and displays the list of labeled projects on a human-computer interaction interface. The list of labeled projects includes multiple labeled projects. The training platform parses the data of each item in the list of labeled projects to obtain the labeled project information; The training platform selects any one of the labeled projects from the human-computer interaction interface; The training platform obtains the corresponding annotation data from the annotation software based on the selected annotation project; The training platform converts the labeled data into a dataset; The training platform will automatically call the dataset and configuration data to train the AI ​​model and obtain the training results; wherein, the training results include the trained model; The training platform automatically creates an inference process based on the trained model or pre-trained model. The training platform creates automated annotation services for the annotation project selected by the user by calling compatible interface methods.

2. The interaction method between annotation software and training platform according to claim 1, characterized in that, The training platform obtains a list of labeled projects from multiple labeled software programs, including: The training platform sends a request for a list to the annotation software, so that the annotation software can obtain the list of annotation projects based on the request. The training platform receives the list of labeled projects sent by the labeling software.

3. The interaction method between annotation software and training platform according to claim 1, characterized in that, The training platform obtains corresponding annotation data from the annotation software based on the selected annotation project, including: The training platform sends a request to the annotation software based on the selected annotation project, so that the annotation software can obtain the annotation data corresponding to the annotation project based on the request. The training platform receives the annotation data sent by the annotation software.

4. The interaction method between annotation software and training platform according to claim 1, characterized in that, The training platform creates automated annotation services for the annotation project selected by the user by calling compatible interface methods, including: The training platform acquires the labeled project selected by the user on the human-computer interaction interface; The training platform configures automated annotation services in the annotation project corresponding to the annotation software by calling the compatible interface method.

5. The interaction method between annotation software and training platform according to claim 4, characterized in that, The method further includes: Once the automated annotation service is successfully configured, the corresponding annotation project can be accessed through the human-computer interaction interface of the annotation software. The annotation software obtains the user's confirmation to start annotation; The annotation software begins automated annotation based on the confirmed annotation information and obtains the annotation results.

6. An interactive system for annotation software and a training platform, characterized in that, The system includes: The annotation project list acquisition module is used by the training platform to acquire an annotation project list from multiple annotation software and display the annotation project list on the human-computer interaction interface. The annotation project list includes multiple annotation projects. The parsing module is used by the training platform to parse the data of each item in the list of labeled projects to obtain the labeled project information; The selection module is used by the training platform to select any one of the annotation projects from the human-computer interaction interface; The annotation data acquisition module is used by the training platform to acquire corresponding annotation data from the annotation software based on the selected annotation project; A conversion module is used by the training platform to convert the labeled data into a dataset; The training module is used by the training platform to automatically call the dataset and configuration data to train the AI ​​model and obtain training results; wherein, the training results include the trained model; An inference process creation module is used by the training platform to automatically create an inference process based on the trained model or pre-trained model. An automated annotation service creation module is used by the training platform to create automated annotation services for the annotation project selected by the user by calling compatible interface methods.

7. The interaction system between the annotation software and the training platform according to claim 6, characterized in that, The module for obtaining the labeled project list is specifically used for: The training platform sends a request for a list to the annotation software, so that the annotation software can obtain the list of annotation projects based on the request. The training platform receives the list of labeled projects sent by the labeling software.

8. The interaction system between the annotation software and the training platform according to claim 6, characterized in that, The labeled data acquisition module is specifically used for: The training platform sends a request to the annotation software based on the selected annotation project, so that the annotation software can obtain the annotation data corresponding to the annotation project based on the request. The training platform receives the annotation data sent by the annotation software.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, characterized in that, When the processor executes the computer program, it implements the method described in any one of claims 1 to 5.

10. A computer-readable medium having processor-executable non-volatile program code, characterized in that, The program code causes the processor to execute the method described in any one of claims 1 to 5.

Images