Visualization processing method and device of image model training system

Abstract

The invention discloses a visualization processing method and device for an image model training system, and the method enables a user to quickly select a preset function on a menu bar through a preset image model training system, the set menu bar and a display region, and responds to a page corresponding to the selected preset function in the display region. Therefore, the user can perform model training operation on the page. According to the invention, through the preset main interface of the image model training system and the operation process, the standard of image model training is standardized, so that a user can quickly carry out image model training development based on the image model training system, the problem of difficult operation in training of a deep learning algorithm model is solved, and the user experience is improved. And the efficiency of image model training is greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of deep learning model technology, and in particular to a visualization processing method and device for an image model training system. Background Art

[0002] In the industrial field, the application of deep learning has become an important means to improve production efficiency and product quality. However, the existing deep learning algorithm model requires users to completely independently design an algorithm for training, and there is no unified training method, which leads to the problem of operational difficulties in the training of deep learning algorithm models. Therefore, an easy-to-operate training platform is needed.

[0003] Therefore, there is an urgent need for a visualization processing strategy for image model training systems to solve the operational difficulties in training deep learning algorithm models. Summary of the invention

[0004] The embodiments of the present invention provide a visualization processing method and device for an image model training system to solve the problem that the training of a deep learning algorithm model has operational difficulties.

[0005] In order to solve the above problem, an embodiment of the present invention provides a visualization processing method of an image model training system, comprising:

[0006] In response to a selected operation performed by a user on an image data management control, an image data management instruction is generated, and an image data management page is opened according to the image data management instruction, so that the user processes training samples in the image data management page;

[0007] In response to a selected operation performed by a user on a model training control, a model training instruction is generated, and a model training page is opened according to the model training instruction, so that the user performs model training based on the training sample in the model training page to generate an image model;

[0008] In response to a selected operation performed by a user on a model evaluation control, a model evaluation instruction is generated, and a model evaluation page is opened according to the model evaluation instruction, so that the user performs a model evaluation based on the image model in the model evaluation page to obtain an evaluation result;

[0009] Among them, the main interface of the image model training system includes: a menu bar and a display area; the menu bar includes the image data management control, the model training control and the model evaluation control; the display area includes: one of the image data management page, the model training page and the model evaluation page.

[0010] As an improvement of the above scheme, it also includes: generating an inference deployment instruction in response to a selected operation performed by a user on an inference deployment control, and opening an inference deployment page according to the inference deployment instruction, so that the user can perform model monitoring based on the final image model in the inference deployment page; wherein the menu bar also includes an inference deployment control; and the display area also includes: an inference deployment page.

[0011] As an improvement of the above solution, the image data management instruction includes: a data upload instruction and a data labeling instruction; the processing of the training sample includes:

[0012] In response to a user's operation on a data upload control in an image data management page, a data upload instruction is generated, and a data upload page is opened according to the data upload instruction, so that the user can upload an image to be trained on the data upload page;

[0013] In response to the user's operation on the annotation tool control in the image data management page, a data annotation instruction is generated, and a data annotation page is opened according to the data annotation instruction, so that the user can annotate the image to be trained on the data annotation page to obtain a training sample.

[0014] As an improvement of the above solution, the model training instruction includes: a model parameter configuration instruction and a training start instruction; the generation of the image model includes:

[0015] In response to the user's selection of a preset model architecture on the model training page, a model parameter configuration instruction is generated, and a model parameter configuration page is opened according to the model parameter configuration instruction, so that the user can configure the model parameters on the model parameter configuration page;

[0016] In response to the user's operation of confirming the training control in the model parameter configuration page, a training start instruction is generated, and according to the training start instruction, the model is trained based on the selected model architecture, the configured model parameters and the training samples to obtain an image model.

[0017] As an improvement of the above solution, the step of performing model evaluation based on the image model to obtain an evaluation result includes:

[0018] Performing model evaluation on the image model based on the validation set to obtain an accuracy curve, a recall curve, a precision-recall curve, a confusion matrix and an F1 curve;

[0019] The accuracy curve, recall curve, precision-recall curve, confusion matrix, and F1 curve are displayed as evaluation results on the model evaluation page.

[0020] Accordingly, an embodiment of the present invention further provides a visualization processing device of an image model training system, comprising: a data management module, a model training module and a model evaluation module;

[0021] The data management module is used to generate an image data management instruction in response to a selected operation performed by a user on the image data management control, and to open an image data management page according to the image data management instruction, so that the user can process training samples in the image data management page;

[0022] The model training module is used to respond to the selected operation performed by the user on the model training control, generate a model training instruction, and open a model training page according to the model training instruction, so that the user performs model training based on the training sample in the model training page to generate an image model;

[0023] The model evaluation module is used to respond to the selected operation performed by the user on the model evaluation control, generate a model evaluation instruction, and open a model evaluation page according to the model evaluation instruction, so that the user can perform model evaluation based on the image model in the model evaluation page to obtain an evaluation result;

[0024] Among them, the main interface of the image model training system includes: a menu bar and a display area; the menu bar includes the image data management control, the model training control and the model evaluation control; the display area includes: one of the image data management page, the model training page and the model evaluation page.

[0025] As an improvement of the above scheme, it also includes: generating an inference deployment instruction in response to a selected operation performed by a user on an inference deployment control, and opening an inference deployment page according to the inference deployment instruction, so that the user can perform model monitoring based on the final image model in the inference deployment page; wherein the menu bar also includes an inference deployment control; and the display area also includes: an inference deployment page.

[0026] As an improvement of the above solution, the image data management instruction includes: a data upload instruction and a data labeling instruction; the processing of the training sample includes:

[0027] In response to a user's operation on a data upload control in an image data management page, a data upload instruction is generated, and a data upload page is opened according to the data upload instruction, so that the user can upload an image to be trained on the data upload page;

[0028] In response to the user's operation on the annotation tool control in the image data management page, a data annotation instruction is generated, and a data annotation page is opened according to the data annotation instruction, so that the user can annotate the image to be trained on the data annotation page to obtain a training sample.

[0029] As an improvement of the above solution, the model training instruction includes: a model parameter configuration instruction and a training start instruction; the generation of the image model includes:

[0030] In response to the user's selection of a preset model architecture on the model training page, a model parameter configuration instruction is generated, and a model parameter configuration page is opened according to the model parameter configuration instruction, so that the user can configure the model parameters on the model parameter configuration page;

[0031] In response to the user's operation of confirming the training control in the model parameter configuration page, a training start instruction is generated, and according to the training start instruction, the model is trained based on the selected model architecture, the configured model parameters and the training samples to obtain an image model.

[0032] As an improvement of the above solution, the step of performing model evaluation based on the image model to obtain an evaluation result includes:

[0033] Performing model evaluation on the image model based on the validation set to obtain an accuracy curve, a recall curve, a precision-recall curve, a confusion matrix and an F1 curve;

[0034] The accuracy curve, recall curve, precision-recall curve, confusion matrix, and F1 curve are displayed as evaluation results on the model evaluation page.

[0035] As can be seen from the above, the present invention has the following beneficial effects:

[0036] The present invention provides a visualization processing method for an image model training system. Through a preset image model training system, a menu bar and a display area are set, so that a user can quickly select a preset function on the menu bar, and a page corresponding to the preset function is selected in response to the display area, so that the user can perform model training operations on the page. The present invention standardizes the standard of image model training through the main interface of the preset image model training system and the operation process, so that the user can quickly perform image model training development based on the image model training system, solves the problem of operational difficulties in the training of deep learning algorithm models, and greatly improves the efficiency of image model training. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a flowchart of a visualization processing method of an image model training system provided by an embodiment of the present invention;

[0038] Figure 2 is a structural schematic diagram of a visualization processing device of an image model training system provided by an embodiment of the present invention;

[0039] Figure 3 It is a schematic diagram of the structure of a terminal device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] Embodiment 1

[0042] See also Figure 1 , Figure 1 FIG. 1 is a flow chart of a visualization processing method of an image model training system provided by an embodiment of the present invention. Figure 1 As shown, this embodiment includes steps 101 to 103, and each step is specifically as follows:

[0043] Step 101: In response to a selected operation performed by a user on an image data management control, an image data management instruction is generated, and an image data management page is opened according to the image data management instruction, so that the user can process training samples in the image data management page.

[0044] In a specific embodiment, the image data management control location and function: In the main interface, the image data management control is located in the "Data Management" option in the left menu bar. The user can click "Data Management" to enter the data import interface. This module supports batch upload function, and the user can easily upload files in batches. After the upload is completed, the user can view the uploaded images through thumbnails, and perform batch deletion, view label assignments, etc., making data management more efficient and convenient.

[0045] It can be understood that this embodiment is easy to operate: through a concise and clear interface design and an efficient operation process, the user can complete the management of a large number of images in a short time, which significantly improves the efficiency of data processing.

[0046] In this embodiment, the image data management instruction includes: a data upload instruction and a data labeling instruction; the processing of the training sample includes:

[0047] In response to a user's operation on a data upload control in an image data management page, a data upload instruction is generated, and a data upload page is opened according to the data upload instruction, so that the user can upload an image to be trained on the data upload page;

[0048] In response to the user's operation on the annotation tool control in the image data management page, a data annotation instruction is generated, and a data annotation page is opened according to the data annotation instruction, so that the user can annotate the image to be trained on the data annotation page to obtain a training sample.

[0049] In a specific embodiment, the position and function of the annotation tool control: the image annotation control is located in the "Annotation Tool" option in the "Data Management" module. The user can select the image to be annotated in this module and enter the annotation interface. The annotation tool supports multiple annotation methods such as rectangle, circle, polygon, etc., and provides automatic annotation and intelligent auxiliary annotation functions. The user can customize the annotation category and attributes to adapt to different application scenarios.

[0050] It is understandable that this embodiment is easy to operate: the annotation interface design is simple and intuitive, and users can quickly complete annotation tasks through shortcut keys and drag operations. The platform also supports real-time save and undo operations to ensure the flexibility and convenience of the annotation process. For large-scale annotation tasks, the platform provides an automatic annotation function, which can greatly improve the annotation efficiency.

[0051] Step 102: In response to a selected operation performed by a user on a model training control, a model training instruction is generated, and a model training page is opened according to the model training instruction, so that the user performs model training based on the training samples in the model training page to generate an image model.

[0052] In this embodiment, the model training instruction includes: a model parameter configuration instruction and a training start instruction; the generation of the image model includes:

[0053] In response to the user's selection of a preset model architecture on the model training page, a model parameter configuration instruction is generated, and a model parameter configuration page is opened according to the model parameter configuration instruction, so that the user can configure the model parameters on the model parameter configuration page;

[0054] In response to the user's operation of confirming the training control in the model parameter configuration page, a training start instruction is generated, and according to the training start instruction, the model is trained based on the selected model architecture, the configured model parameters and the training samples to obtain an image model.

[0055] In a specific embodiment, the model training control location and function: the model training module is located in the "Model Training" option in the menu bar on the left side of the main interface. Users can select a predefined model or a custom model architecture in this module and load training data. The platform provides a detailed parameter configuration interface, and users can adjust parameters such as learning rate, batch size, and number of training rounds as needed.

[0056] It is understandable that this embodiment is easy to operate: the platform provides automated training scripts and preset templates, and users can quickly start the training process. During the training process, users can monitor the training progress in real time and view the loss and accuracy changes of the model through visual charts. Users can flexibly adjust the training plan to improve training efficiency.

[0057] Step 103: In response to the selected operation performed by the user on the model evaluation control, a model evaluation instruction is generated, and a model evaluation page is opened according to the model evaluation instruction, so that the user performs model evaluation based on the image model in the model evaluation page to obtain an evaluation result;

[0058] Among them, the main interface of the image model training system includes: a menu bar and a display area; the menu bar includes the image data management control, the model training control and the model evaluation control; the display area includes: one of the image data management page, the model training page and the model evaluation page.

[0059] In this embodiment, the performing model evaluation based on the image model to obtain an evaluation result includes:

[0060] Performing model evaluation on the image model based on the validation set to obtain an accuracy curve, a recall curve, a precision-recall curve, a confusion matrix and an F1 curve;

[0061] The accuracy curve, recall curve, precision-recall curve, confusion matrix, and F1 curve are displayed as evaluation results on the model evaluation page.

[0062] It should be noted that the F1 curve shows the overall performance of the model. The precision-recall (PR) curve reflects the detection effect of the model at different thresholds. The precision (P) curve and the recall (R) curve respectively show the accuracy and coverage of the model in detection. The confusion matrix visualizes the classification accuracy between different categories.

[0063] In a specific embodiment, the evaluation result includes: accuracy, recall rate and F1 score; and determining the final image model based on the evaluation result includes:

[0064] The evaluation result corresponding to the image model is used as the evaluation input, and the image model is used as the target model. The evaluation operation is repeated until the final image model is generated.

[0065] The evaluation operation is specifically as follows:

[0066] Based on the evaluation input, make a judgment on the target model;

[0067] If the accuracy is greater than or equal to the accuracy threshold, the recall is greater than or equal to the recall threshold, and the F1 score is greater than or equal to the F1 score threshold, the initial image model is used as the final image model;

[0068] Otherwise, the model parameter adjustment page corresponding to the initial image model pops up, responds to the user's parameter modification of the initial image model in the model parameter adjustment page, obtains the transition image model, evaluates the transition image model, generates a transition evaluation result, uses the transition evaluation result as the evaluation input, uses the transition image model as the target model, and re-executes the evaluation operation.

[0069] In a specific embodiment, the location and function of the model evaluation control: The model evaluation control is located in the "Model Evaluation" option in the menu bar on the left side of the main interface. Users can select a trained model in this module for evaluation testing. The platform provides a variety of evaluation indicators, such as accuracy, recall, F1 score, etc., and displays the evaluation results in charts.

[0070] It can be understood that this embodiment is easy to operate: through an intuitive evaluation interface, the user can quickly understand the performance of the model and optimize the model according to the evaluation results.

[0071] In this embodiment, it also includes: generating an inference deployment instruction in response to a selected operation performed by a user on an inference deployment control, and opening an inference deployment page according to the inference deployment instruction, so that the user can perform model monitoring based on the final image model in the inference deployment page; wherein the menu bar also includes an inference deployment control; and the display area also includes: an inference deployment page.

[0072] In a specific embodiment, the location and function of the inference deployment control: The inference deployment control is located in the "Inference Deployment" option in the menu bar on the left side of the main interface. Users can select a trained model in this module, configure deployment parameters, and select a deployment environment. The platform supports local deployment, cloud deployment, and edge device deployment, and users can choose the appropriate deployment method according to actual needs.

[0073] It is understandable that this embodiment is easy to operate: the deployment interface is simple in design, and users can complete the deployment of the model in just a few steps. The platform provides a detailed deployment guide and real-time monitoring function, and users can check the deployment status and inference results at any time to ensure the smooth progress of the deployment process.

[0074] See also Figure 2 , Figure 2 2 is a schematic diagram of a structure of a visualization processing device of an image model training system provided by an embodiment of the present invention, comprising: a data management module 201, a model training module 202 and a model evaluation module 203;

[0075] The data management module is used to generate an image data management instruction in response to a selected operation performed by a user on the image data management control, and to open an image data management page according to the image data management instruction, so that the user can process training samples in the image data management page;

[0076] The model training module is used to respond to the selected operation performed by the user on the model training control, generate a model training instruction, and open a model training page according to the model training instruction, so that the user performs model training based on the training sample in the model training page to generate an image model;

[0077] The model evaluation module is used to respond to the selected operation performed by the user on the model evaluation control, generate a model evaluation instruction, and open a model evaluation page according to the model evaluation instruction, so that the user can perform model evaluation based on the image model in the model evaluation page to obtain an evaluation result;

[0078] Among them, the main interface of the image model training system includes: a menu bar and a display area; the menu bar includes the image data management control, the model training control and the model evaluation control; the display area includes: one of the image data management page, the model training page and the model evaluation page.

[0079] As an improvement of the above scheme, it also includes: generating an inference deployment instruction in response to a selected operation performed by a user on an inference deployment control, and opening an inference deployment page according to the inference deployment instruction, so that the user can perform model monitoring based on the final image model in the inference deployment page; wherein the menu bar also includes an inference deployment control; and the display area also includes: an inference deployment page.

[0080] As an improvement of the above solution, the image data management instruction includes: a data upload instruction and a data labeling instruction; the processing of the training sample includes:

[0081] In response to a user's operation on a data upload control in an image data management page, a data upload instruction is generated, and a data upload page is opened according to the data upload instruction, so that the user can upload an image to be trained on the data upload page;

[0082] In response to the user's operation on the annotation tool control in the image data management page, a data annotation instruction is generated, and a data annotation page is opened according to the data annotation instruction, so that the user can annotate the image to be trained on the data annotation page to obtain a training sample.

[0083] As an improvement of the above solution, the model training instruction includes: a model parameter configuration instruction and a training start instruction; the generation of the image model includes:

[0084] In response to the user's selection of a preset model architecture on the model training page, a model parameter configuration instruction is generated, and a model parameter configuration page is opened according to the model parameter configuration instruction, so that the user can configure the model parameters on the model parameter configuration page;

[0085] In response to the user's operation of confirming the training control in the model parameter configuration page, a training start instruction is generated, and according to the training start instruction, the model is trained based on the selected model architecture, the configured model parameters and the training samples to obtain an image model.

[0086] As an improvement of the above solution, the step of performing model evaluation based on the image model to obtain an evaluation result includes:

[0087] Performing model evaluation on the image model based on the validation set to obtain an accuracy curve, a recall curve, a precision-recall curve, a confusion matrix and an F1 curve;

[0088] The accuracy curve, recall curve, precision-recall curve, confusion matrix, and F1 curve are displayed as evaluation results on the model evaluation page.

[0089] This embodiment uses a preset image model training system, a set menu bar and a display area, so that the user can quickly select a preset function on the menu bar, and responds to the page corresponding to the preset function in the display area, so that the user can perform model training operations on the page. The present invention standardizes the standard of image model training through the main interface of the preset image model training system and the operation process, so that the user can quickly perform image model training development based on the image model training system, solves the problem of operational difficulties in the training of deep learning algorithm models, and greatly improves the efficiency of image model training.

[0090] Embodiment 2

[0091] See also Figure 3 , Figure 3 It is a schematic diagram of the structure of a terminal device provided in one embodiment of the present invention.

[0092] A terminal device of this embodiment includes: a processor 301, a memory 302, and a computer program stored in the memory 302 and executable on the processor 301. When the processor 301 executes the computer program, the steps of the visualization processing method of the above-mentioned image model training system in the embodiment are implemented, for example Figure 1 Alternatively, when the processor executes the computer program, the functions of each module in the above-mentioned device embodiments are realized, for example: Figure 2All modules of the visualization processing device of the image model training system are shown.

[0093] In addition, an embodiment of the present invention also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the visualization processing method of the image model training system described in any of the above embodiments.

[0094] Those skilled in the art will understand that the schematic diagram is merely an example of a terminal device and does not constitute a limitation on the terminal device. The terminal device may include more or fewer components than shown in the diagram, or a combination of certain components, or different components. For example, the terminal device may also include input and output devices, network access devices, buses, etc.

[0095] The processor 301 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. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor 301 is the control center of the terminal device, and uses various interfaces and lines to connect various parts of the entire terminal device.

[0096] The memory 302 can be used to store the computer program and / or module. The processor 301 implements various functions of the terminal device by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory 302. The memory 302 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0097] Wherein, if the module / unit integrated in the terminal device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0098] It should be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0099] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A visualization processing method for an image model training system, characterized in that: include: In response to a selected operation performed by a user on an image data management control, an image data management instruction is generated, and an image data management page is opened according to the image data management instruction, so that the user processes training samples in the image data management page; In response to a selected operation performed by a user on a model training control, a model training instruction is generated, and a model training page is opened according to the model training instruction, so that the user performs model training based on the training sample in the model training page to generate an image model; In response to a selected operation performed by a user on a model evaluation control, a model evaluation instruction is generated, and a model evaluation page is opened according to the model evaluation instruction, so that the user performs a model evaluation based on the image model in the model evaluation page to obtain an evaluation result; Among them, the main interface of the image model training system includes: a menu bar and a display area; the menu bar includes the image data management control, the model training control and the model evaluation control; the display area includes: one of the image data management page, the model training page and the model evaluation page.

2. The visualization processing method of the image model training system according to claim 1, characterized in that: Also includes: In response to the selected operation performed by the user on the inference deployment control, an inference deployment instruction is generated, and an inference deployment page is opened according to the inference deployment instruction, so that the user can perform model monitoring based on the final image model in the inference deployment page; wherein the menu bar also includes the inference deployment control; the display area also includes: the inference deployment page.

3. The visualization processing method of the image model training system according to claim 2, characterized in that: The image data management instructions include: data upload instructions and data labeling instructions; the processing of the training samples includes: In response to a user's operation on a data upload control in an image data management page, a data upload instruction is generated, and a data upload page is opened according to the data upload instruction, so that the user can upload an image to be trained on the data upload page; In response to the user's operation on the annotation tool control in the image data management page, a data annotation instruction is generated, and a data annotation page is opened according to the data annotation instruction, so that the user can annotate the image to be trained on the data annotation page to obtain a training sample.

4. The visualization processing method of the image model training system according to claim 2, characterized in that: The model training instruction includes: a model parameter configuration instruction and a training start instruction; the generation of the image model includes: In response to the user's selection of a preset model architecture on the model training page, a model parameter configuration instruction is generated, and a model parameter configuration page is opened according to the model parameter configuration instruction, so that the user can configure the model parameters on the model parameter configuration page; In response to the user's operation of confirming the training control in the model parameter configuration page, a training start instruction is generated, and according to the training start instruction, the model is trained based on the selected model architecture, the configured model parameters and the training samples to obtain an image model.

5. The visualization processing method of the image model training system according to claim 4, characterized in that: The performing model evaluation based on the image model to obtain an evaluation result includes: Performing model evaluation on the image model based on the validation set to obtain an accuracy curve, a recall curve, a precision-recall curve, a confusion matrix and an F1 curve; The accuracy curve, recall curve, precision-recall curve, confusion matrix, and F1 curve are displayed as evaluation results on the model evaluation page.

6. A visualization processing device for an image model training system, characterized in that: include: Data management module, model training module and model evaluation module; The data management module is used to generate an image data management instruction in response to a selected operation performed by a user on the image data management control, and to open an image data management page according to the image data management instruction, so that the user can process training samples in the image data management page; The model training module is used to respond to the selected operation performed by the user on the model training control, generate a model training instruction, and open a model training page according to the model training instruction, so that the user performs model training based on the training sample in the model training page to generate an image model; The model evaluation module is used to respond to the selected operation performed by the user on the model evaluation control, generate a model evaluation instruction, and open a model evaluation page according to the model evaluation instruction, so that the user can perform model evaluation based on the image model in the model evaluation page to obtain an evaluation result; Among them, the main interface of the image model training system includes: a menu bar and a display area; the menu bar includes the image data management control, the model training control and the model evaluation control; the display area includes: one of the image data management page, the model training page and the model evaluation page.

7. The visualization processing device of the image model training system according to claim 6, characterized in that: Also includes: In response to the selected operation performed by the user on the inference deployment control, an inference deployment instruction is generated, and an inference deployment page is opened according to the inference deployment instruction, so that the user can perform model monitoring based on the final image model in the inference deployment page; wherein the menu bar also includes the inference deployment control; the display area also includes: the inference deployment page.

8. The visualization processing device of the image model training system according to claim 7, characterized in that: The image data management instructions include: data upload instructions and data labeling instructions; the processing of the training samples includes: In response to a user's operation on a data upload control in an image data management page, a data upload instruction is generated, and a data upload page is opened according to the data upload instruction, so that the user can upload an image to be trained on the data upload page; In response to the user's operation on the annotation tool control in the image data management page, a data annotation instruction is generated, and a data annotation page is opened according to the data annotation instruction, so that the user can annotate the image to be trained on the data annotation page to obtain a training sample.

9. The visualization processing device of the image model training system according to claim 7, characterized in that: The model training instruction includes: a model parameter configuration instruction and a training start instruction; the generation of the image model includes: In response to the user's selection of a preset model architecture on the model training page, a model parameter configuration instruction is generated, and a model parameter configuration page is opened according to the model parameter configuration instruction, so that the user can configure the model parameters on the model parameter configuration page; In response to the user's operation of confirming the training control in the model parameter configuration page, a training start instruction is generated, and according to the training start instruction, the model is trained based on the selected model architecture, the configured model parameters and the training samples to obtain an image model.

10. The visualization processing device of the image model training system according to claim 9, characterized in that: The performing model evaluation based on the image model to obtain an evaluation result includes: Performing model evaluation on the image model based on the validation set to obtain an accuracy curve, a recall curve, a precision-recall curve, a confusion matrix and an F1 curve; The accuracy curve, recall curve, precision-recall curve, confusion matrix, and F1 curve are displayed as evaluation results on the model evaluation page.

Images