CN117763355A - Iterative training system and method for industrial quality inspection model and industrial quality inspection system - Google Patents
Iterative training system and method for industrial quality inspection model and industrial quality inspection system Download PDFInfo
- Publication number
- CN117763355A CN117763355A CN202311776484.4A CN202311776484A CN117763355A CN 117763355 A CN117763355 A CN 117763355A CN 202311776484 A CN202311776484 A CN 202311776484A CN 117763355 A CN117763355 A CN 117763355A
- Authority
- CN
- China
- Prior art keywords
- model
- quality inspection
- data
- defect
- user
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000012549 training Methods 0.000 title claims abstract description 123
- 238000007689 inspection Methods 0.000 title claims abstract description 115
- 238000000034 method Methods 0.000 title claims abstract description 34
- 230000007547 defect Effects 0.000 claims abstract description 136
- 230000000007 visual effect Effects 0.000 claims abstract description 41
- 238000002372 labelling Methods 0.000 claims abstract description 39
- 238000013480 data collection Methods 0.000 claims abstract description 11
- 238000012360 testing method Methods 0.000 claims description 42
- 230000003993 interaction Effects 0.000 claims description 21
- 230000002950 deficient Effects 0.000 claims description 20
- 238000013515 script Methods 0.000 claims description 10
- 238000004891 communication Methods 0.000 claims description 5
- 230000002452 interceptive effect Effects 0.000 claims description 4
- 238000004422 calculation algorithm Methods 0.000 description 13
- 238000004519 manufacturing process Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 5
- 238000013473 artificial intelligence Methods 0.000 description 3
- 238000012372 quality testing Methods 0.000 description 3
- 238000012217 deletion Methods 0.000 description 2
- 230000037430 deletion Effects 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses an iterative training system and method of an industrial quality inspection model and the industrial quality inspection system, wherein the iterative training system comprises a data collection module, a data labeling module and a model iteration module, and the data collection module stores defect data into a defect product database of the iterative training system; the data labeling module is used for providing a visual interface to display defect data in the defect product database, guiding a user to select the defect data to carry out online data labeling through visual information, and storing the defect data labeled by the user as a model iteration sample into a model iteration data set; the model iteration module is used for a user to select an industrial quality inspection model to be updated, and sends a target industrial quality inspection model selected by the user and a model iteration sample in a model iteration data set to the server equipment for model iteration training. The method and the device can enable industrial workers without computer knowledge to also collect the defect data to iterate the industrial quality inspection model, and reduce cost.
Description
Technical Field
The application relates to the technical field of industrial artificial intelligence, in particular to an iterative training system and method of an industrial quality inspection model and an industrial quality inspection system.
Background
With the continuous development of artificial intelligence technology, the application range of artificial intelligence models is also becoming wider and wider. In actual production, the products produced on the production line often have model and parameter changes, so that the existing models are easy to have a large number of errors in recognition, and further production is blocked. Therefore, the algorithm model needs to be iteratively updated continuously, so that the algorithm model is better suitable for the current application scene.
Since most industrial workers do not have sufficient computer knowledge and are more unaware of the computer algorithm, the computer company or the factory is required to recruit the algorithm engineer specially and provide an image processor (Graphics Processing Unit, GPU), relevant data of a new product collected in advance is submitted to the algorithm engineer, the algorithm engineer designs a model, the model is correspondingly trained by utilizing the collected relevant data, and the trained model is submitted to a production line server for use.
However, this method in the related art is very high in both communication cost, hardware cost and labor cost, and particularly, when a server is shared by a plurality of production lines and a plurality of factories, a large amount of cost is required to be input.
Disclosure of Invention
In view of the above problems, the present application provides an iterative training system and method for an industrial quality inspection model, and an industrial quality inspection system, so as to solve the above technical problems.
In a first aspect, the present application provides an iterative training system for an industrial quality inspection model, applied to a client device communicatively connected to a server device, the iterative training system for an industrial quality inspection model comprising:
the pre-configuration module is used for providing interactive information for the iterative training system so that a user can operate the iterative training system according to the interactive information;
the data collection module is used for acquiring defect data of the defect products to be inspected through quality inspection equipment arranged in the industrial quality inspection environment and storing the defect data into a defect product database of the iterative training system;
the data labeling module is used for providing a visual interface to display defect data in the defect product database, guiding a user to select the defect data to carry out online data labeling through visual information, and storing the defect data labeled by the user as a model iteration sample into a model iteration data set;
the model iteration module is used for a user to select an industrial quality inspection model to be updated, and sending the target industrial quality inspection model selected by the user and model iteration samples in the model iteration data set to the server-side equipment so as to call training resources through the server-side equipment and utilize the model iteration samples in the model iteration data set to carry out iteration training on the target industrial quality inspection model.
In one possible implementation manner of the present application, the defect data is a photograph of the defect product, and the quality inspection device is a camera;
the data labeling module is used for:
displaying a photo of the defective product through a visual interface;
displaying marking guide information through a visual interface to guide a user to mark a defect area of a defect product in the displayed photo;
and generating a labeling file comprising labeling information of the defective product based on the labeling result of the photo by the user.
In one possible implementation manner of the present application, the labeling file includes product label information of the defective product and endpoint information of a labeling frame that labels the defective area.
In one possible implementation manner of the present application, the iterative training system further includes:
the model test module is used for a user to select an industrial quality inspection model to be tested, and sends the test sample in the test data set of the model to be tested and the iterative training system selected by the user to the server equipment so as to call the test resource through the server equipment to test the model to be tested by utilizing the test sample in the test data set, and a test record comprising a test result is generated.
In one possible implementation of the present application, the interaction information includes one or more information files in a running environment, a running script, a description, and a manual.
In one possible implementation manner of the present application, the pre-configuration module is configured to provide independent rights to an administrator, and when the administrator opens the independent rights, the iterative training system does not respond to a download instruction or an update instruction proposed by a user other than the administrator.
In one possible implementation manner of the application, the pre-configuration module is further used for providing a path for displaying the interaction information by the visual interface.
In one possible implementation manner of the application, the pre-configuration module is further configured to provide corresponding interaction information to the user according to the received call instruction, where the call instruction carries a path of the interaction information.
In a second aspect, the present application further provides an iterative training method of an industrial quality inspection model, applied to a client device communicatively connected to a server device, where the iterative training method of the industrial quality inspection model includes:
acquiring defect data of a defect product to be inspected through quality inspection equipment arranged in an industrial quality inspection environment, and storing the defect data into a defect product database;
providing a visual interface to display defect data in a defect product database, guiding a user to select the defect data to carry out online data annotation through visual information, and storing the defect data marked by the user as a model iteration sample into a model iteration data set;
and the user selects the industrial quality inspection model to be updated, and the target industrial quality inspection model selected by the user and the model iteration sample in the model iteration data set are sent to the server-side equipment, so that the server-side equipment invokes the training resource to utilize the model iteration sample in the model iteration data set to carry out iteration training on the target industrial quality inspection model.
In a third aspect, the present application further provides an industrial quality inspection system, including a client device and a server device that are communicatively connected;
the client device is used for acquiring the defect data of the defect product to be inspected through the quality inspection device arranged in the industrial quality inspection environment and storing the defect data into a defect product database of the iterative training system,
providing a visual interface to display defect data in a defect product database, guiding a user to select the defect data to carry out online data marking through visual information, storing the defect data marked by the user as a model iteration sample into a model iteration data set,
the method comprises the steps that a user selects an industrial quality inspection model to be updated, and a target industrial quality inspection model selected by the user and a model iteration sample in a model iteration data set are sent to a server device;
and the server equipment is used for calling the model iteration sample in the training resource utilization model iteration data set to carry out iteration training on the target industrial quality inspection model.
From the above, the present application has the following advantages:
according to the method and the device, interaction between a user and an engineer is achieved through the pre-configuration module, defect data of a defect product are stored in a defect product database through the data collection module, a visual interface is provided for displaying the defect data through the data marking module, the user selects the defect data to conduct online data marking through visual information, the marked defect data are stored into a model iteration data set as model iteration samples, then the model iteration module sends a target industrial quality inspection model selected by the user and the model iteration samples in the model iteration data set to the server equipment, the target industrial quality inspection model is subjected to iteration training through the server equipment by utilizing the model iteration samples, the automation degree is high, even workers without computer knowledge can automatically collect the defect data to conduct iteration of the industrial quality inspection model, and the cost is reduced.
These and other aspects of the present application will be more readily apparent from the following description of the embodiments.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the description of the embodiments will be briefly introduced below, it being obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic view of one scenario of an iterative training system provided in an embodiment of the present application;
FIG. 2 is a block diagram of an iterative training system provided in an embodiment of the present application;
FIG. 3 is a schematic illustration of an annotation of defect data provided in an embodiment of the present application;
FIG. 4 is an interface schematic of an annotation interface provided in an embodiment of the present application;
FIG. 5 is another block diagram of an iterative training system provided in an embodiment of the present application;
FIG. 6 is a schematic diagram of an interface for training records provided in embodiments of the present application;
FIG. 7 is a schematic diagram of an interface layout of a visual interface provided in an embodiment of the present application;
FIG. 8 is a flow chart of an iterative training method provided in an embodiment of the present application;
fig. 9 is a schematic structural view of an industrial quality inspection system provided in an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all, of the embodiments of the present application. All other embodiments, which can be made by those skilled in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.
In the description of the present application, it should be understood that the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or an implicit indication of the number of technical features being indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include one or more features. In the description of the present application, the meaning of "a plurality" is two or more, unless explicitly defined otherwise.
In this application, the term "exemplary" is used to mean "serving as an example, instance, or illustration. Any embodiment described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments. The following description is presented to enable any person skilled in the art to make and use the application. In the following description, details are set forth for purposes of explanation. It will be apparent to one of ordinary skill in the art that the present application may be practiced without these specific details. In other instances, well-known structures and processes have not been shown in detail to avoid obscuring the description of the present application with unnecessary detail. Thus, the present application is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed herein.
Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
The embodiment of the application provides an iterative training system and method of an industrial quality inspection model and the industrial quality inspection system, and the iterative training system and the industrial quality inspection system are respectively described in detail below.
First, an embodiment of the present application provides an iterative training system for an industrial quality inspection model, referring to fig. 1, fig. 1 is a schematic view of a scenario of the iterative training system provided in the embodiment of the present application, where the iterative training system is applied to a client device 200 communicatively connected to a server device 100.
In this application, network communication between the server device 100 and the client device 200 may be implemented by any communication means, including, but not limited to, mobile communication based on the third generation partnership project (3rd Generation Partnership Project,3GPP), long term evolution (Long Term Evolution, LTE), worldwide interoperability for microwave access (Worldwide Interoperability for Microwave Access, wiMAX), or computer network communication based on the TCP/IP protocol family (TCP/IP Protocol Suite, TCP/IP), user datagram protocol (User Datagram Protocol, UDP), and so on.
The server device 100 may be a stand-alone server, or may be a server network or a server cluster formed by servers, for example, the server device 100 described in the present application includes, but is not limited to, a computer, a network host, a single network server, a plurality of network server sets, or a cloud server formed by a plurality of servers. Wherein the Cloud server is composed of a large number of computers or web servers based on Cloud Computing (Cloud Computing).
Client device 200 may be a general purpose computer device or a special purpose computer device. Client device 200 may be an industrial computer, or the like in a particular implementation.
In this embodiment of the present application, the client device 200 may be further communicatively connected to a quality inspection device 300 located on a production line or in an industrial quality inspection environment, where the quality inspection device 300 may collect product data or information of a product to be inspected on the production line or in the industrial quality inspection environment.
As shown in fig. 2, fig. 2 is a block diagram of an iterative training system provided in an embodiment of the present application, the iterative training system 400 may include a pre-configuration module 480, a data collection module 410, a data annotation module 420, and a model iteration module 430.
Wherein the pre-configuration module 480 may be configured to provide interaction information for the iterative training system 400 to enable a user to run the iterative training system 400 based on the interaction information.
The data collection module 410 may be configured to obtain defect data for a defective product being inspected by the inspection device 300 disposed in an industrial inspection environment and store the defect data in a defect product database of the iterative training system 400.
The data labeling module 420 may be configured to provide a visual interface to display defect data in the defect product database, guide a user to select the defect data to perform online data labeling through visual information, and store the defect data labeled by the user as a model iteration sample in a model iteration data set.
The model iteration module 430 may be configured to allow a user to select an industrial quality inspection model to be updated, and send the target industrial quality inspection model selected by the user and the model iteration sample in the model iteration data set to the server device 100, so as to invoke the training resource by the server device 100 to iteratively train the target industrial quality inspection model by using the model iteration sample in the model iteration data set.
In this embodiment of the present application, the interaction information may be an information file such as an operation environment, an operation script, a description, a manual, or the like, which is uploaded by an algorithm engineer when building a model, and when a user performs iterative training on an industrial quality inspection module by operating the iterative training system, the iterative training system may be operated based on the foregoing interaction information.
In an application scenario of performing iterative training on an industrial quality inspection model, the quality inspection device 300 is disposed in an industrial quality inspection environment, and may be used to obtain defect data of a defective product, where the defect data may be image data, sound data or electrical signal data, and may specifically be determined according to an actual application scenario, which is not limited herein.
After the data collection module 410 obtains the defect data of the defect product through the quality inspection device 300, the defect data may be transmitted to a defect product database for storage.
In some embodiments, the data collection module 410 may transmit the defect data to the defect product database immediately after each acquisition of the defect data.
In other embodiments, the data collection module 410 may also be configured with a buffer unit, through which the defect data from the quality inspection apparatus 300 is buffered, and then the buffered defect data is transferred to the defect product database for storage.
In this embodiment, the data labeling module 420 may display the defect data stored in the defect product database to the user through the visual interface, and guide the user to select and label the defect data through the visual information.
It will be appreciated that the user may select defect data based on the visual interface. For example, the data labeling module 420 may provide product classification, acquisition time of defect data, etc. on a screen, the user may select defect data to be labeled based on the product classification, the acquisition time, etc., and after the user selects corresponding defect data, the data labeling module 420 may display the defect data selected by the user.
After the defect data is displayed on the screen, the data labeling module 420 may guide the user to label the defect data displayed on the screen through the prompt information, and it may be understood that the label may be that the user selects a region of interest, for example, a region where the defect is located, and the defect data labeled by the user may be transmitted as a model iteration sample to the model iteration data set for storage.
In this embodiment of the present application, the model iteration module 430 may provide a visual interface for the user to select an industrial quality inspection model that needs to be updated this time, and after the user selects, send the target industrial quality inspection model selected by the user and the model iteration sample corresponding to the target industrial quality inspection model stored in the model iteration dataset to the server device 100.
After receiving the target industrial quality inspection model and the model iteration sample, the server device 100 may invoke a corresponding training resource to utilize the model iteration sample to perform iterative training on the target industrial quality inspection model, so as to update the target industrial quality inspection model.
In this embodiment of the present application, interaction between a user and an algorithm engineer is implemented through the pre-configuration module 480, defect data of a defect product is stored in a defect product database through the data collection module 410, a visual interface is provided by the data labeling module 420 to display the defect data, the user selects the defect data to perform online data labeling through visual information, the labeled defect data is stored as a model iteration sample in a model iteration data set, and then the model iteration module 430 sends a target industrial quality inspection model selected by the user and a model iteration sample in the model iteration data set to the server device 100, so that the server device 100 can utilize the model iteration sample to perform iterative training on the target industrial quality inspection model.
Next, a detailed description of the modules shown in fig. 2 and the specific embodiments that may be employed in the practical application will be continued.
In some embodiments of the present application, the model iteration module 430 may also be configured to receive an industrial quality inspection model sent by the server device 100 after performing the iterative training, for performing industrial quality inspection on a product to be inspected.
In this embodiment of the present application, after the training of the industrial quality inspection model by the server device 100 is completed, the industrial quality inspection model may be stored, or the trained industrial quality inspection model may also be sent to the model iteration module 430, and after the model iteration module 430 receives the trained industrial quality inspection model, the quality inspection model may be invoked to inspect the quality of the product to be inspected in the quality inspection process.
In some embodiments of the present application, the defect data may be a photograph of the defective product, and the quality inspection device 300 may be cameras disposed on a production line, where it may be understood that the number of cameras on the production line may be 1 or more, and may be specifically determined according to an actual application scenario, which is not limited herein.
The data annotation module 420 may also be specifically configured to:
displaying a photo of the defective product through a visual interface; displaying marking guide information through a visual interface to guide a user to mark a defect area of a defect product in the displayed photo; and generating a labeling file comprising labeling information of the defective product based on the labeling result of the photo by the user.
In this embodiment, the photos of the defective products stored in the defective product database may be configured with parameters such as product type, product name, shooting time, etc., and the data labeling module 420 may provide options on the screen interface for the user to search and select the photos in the defective product database according to the parameters, and display the photos on the interface after the user selects the photos.
And then displaying marking guide information on the interface to guide the user to mark the defect area in the photo, for example, as shown in fig. 3, the user is guided to use a marking tool Labelimg to indicate the area where the defect is located in a manner of drawing a rectangular frame.
After labeling is complete, the data labeling module 420 may also direct the user to name the labeled rectangular box and then generate a labeling file that includes labeling information for the defective product.
In one specific implementation, the annotation file may include product label information of the defective product and endpoint information of an annotation box that annotates the defective area.
In this embodiment, the labeling file may be an xml file, and the product tag information may be a name, a type, etc. of a product, so that the xml file may include the name of the labeled product and information of each endpoint of the labeling frame.
It will be appreciated that for a rectangular frame, two end points on the diagonal of the rectangular frame may describe the position information of the entire rectangular frame, and thus, the end point information may be the coordinates of the upper left end point and the lower right end point of the rectangular frame, or the coordinates of the lower left end point and the upper right end point of the rectangular frame.
It should be noted that, if the adopted labeling frame is an irregular opposite frame, the endpoint information may be coordinate information of each inflection point or turning point; if the adopted marked frame is a circular frame, the corresponding endpoint information can be the center coordinate information and the radius of the circular frame.
As an implementation manner, the client device of the present application may be equipped with a file system, which may be an operation interface similar to windows folders, supporting creation, opening, deletion, renaming of folders, and uploading and downloading of files/folders.
Each dataset record has an independent directory under which file operations for that dataset can only be performed.
In the file system, when the mouse selects a picture (the detection format is jpg/png, etc.), a labeling button appears, and clicking the labeling button enters a labeling interface shown in fig. 4, and a save, last/next, and mouse wheel zoom-in/zoom-out button may be arranged on the labeling interface.
It can be appreciated that in some application scenarios, a user may also implement annotation by using an offline program, and only upload the annotated file to the dataset after the annotation is completed.
As shown in fig. 5, in some embodiments of the present application, the iterative training system 400 may further include a training recording module 440, where the training recording module 440 may be configured to record training time, training status, and training results of each training of the industrial quality inspection model by the user, and generate a training record list for the user to view.
It will be appreciated that the use and functionality of the different industrial quality inspection models are different, and that the engineers developing the models may initially define and describe the models, i.e. algorithms, as required and configure them with corresponding parameters, and in subsequent iterative training, may run the training scripts with predefined running commands and predefined parameters or temporarily configured parameters. The run command is a shell command in a user-defined centos 7-based run environment for invoking an executable file.
The configuration is predefined by the user or is filled in when the user clicks "training", the configuration is integrally formed into data in json format, the data comprises multiple parameters, and the data can further comprise list parameters of at most one layer, for example, the first layer of parameters can be list types, and parameters such as a data set, a character string and the like can be added in order inside the list type parameters.
When training is executed, after a user clicks a training button, a predefined operation command is operated in a shell environment of the background, and configured parameters are transmitted in a json file form, so that the training script can be started to be executed.
It may be appreciated that the iterative script may be executed in other operating environments, and specifically may be determined according to an actual application scenario, which is not limited herein.
As shown in fig. 6, the training record may be used to record each training result, which may be a plurality of files. After each training, a record can be added in the training record list on the interface, training model information and states are displayed, a successful training record is displayed, and the file system opening operation can be directly invoked by clicking the training result field or the view button to view detailed training results.
In the file system of opening "training result", the file is selected, a button for adding as model "can be set, the button is clicked, the form is popped up, and after filling in the information, the file can be added to the model page for testing/downloading.
With continued reference to fig. 5, in some embodiments of the present application, the iterative training system 400 may further include a model testing module 450, where the model testing module 450 may be configured to allow a user to select an industrial quality testing model to be tested, and send the user-selected model to be tested and a test sample in a test data set of the iterative training system to the server device 100, so as to invoke, by the server device 100, a test resource to test the model to be tested using the test sample in the test data set, and generate a test record including a test result.
It will be appreciated that model testing is similar to model iterative training, i.e., algorithmic training, essentially by running commands and configurations to execute scripts. Similar to the "algorithm training," the model test module 450 may be used to describe and predefine a test flow, click on the "test" button, and invoke a test script by running commands and incoming parameters. The test record may be used to record each test result, which may be multiple files.
After the test is finished, a record can be added in the test record list to display the information and the state of the test, the record which runs successfully, and the file system opening operation can be directly invoked by clicking the test result field or the view button to view the detailed test result.
In some embodiments of the present application, the iterative training system 400 may further include a rights management module 460 and a logging module 470, where the rights management module 460 may be configured to perform rights management on different users, so that the different users have different operation rights on the iterative training system; the logging module 470 may be used to log the user's operational behavior of various data stored in the iterative training system.
In the interface layout shown in fig. 7, in this embodiment, the left part of each page of the visual interface may be a function menu, the clicking module may expand the sub-module of the module, the clicking sub-module may display the list page on the right, and the operation button of each line of data may be on the rightmost side of the line.
Each page may also support paging (10 by default for a page) and searching (with the additional addition of time search options if there are time fields), searching using fuzzy searching, all fields presented by the interface can be fuzzy searched.
In one particular implementation, menu options may include "user management," enterprise management, "" project management, "" behavior log.
Users can be classified into four types of "supermanager", "enterprise manager", "project manager" and "project user", and users are created by the previous level user without registration.
Wherein, the super manager has all rights of all data, and the system is initially created, and only one is not opened to the outside.
An "enterprise administrator" is created by a "super administrator," which is a one-to-many relationship with an "enterprise," which has all the rights to associate all the data under the enterprise.
The "project manager" is created by the "enterprise manager" and is in one-to-many relationship with the "project" and has all rights to associate all data under the project.
The project user is created by a project manager, is in a many-to-one relationship with the project, and has no deletion authority on all data under the project to which the project user belongs.
Any user cannot delete himself and the same level of users as himself.
In this embodiment, different buttons may be added to the upper right of the list page according to the user category.
For example, "super administrator" corresponds to a newly added "create business" button, "business administrator" corresponds to a newly added "create project" button, and "project administrator" corresponds to a newly added "create project user" button.
The information needed to create the user may include a user name (not necessarily filled), a password (not necessarily filled), a cell phone number (not necessarily filled), a mailbox (not necessarily filled), and the like.
The enterprise is created by a super administrator, and when a user creates the enterprise, the enterprise administrator is automatically generated.
The project is created by an enterprise, and a project manager is automatically generated when the project is created, wherein the project and the enterprise are in a many-to-one relationship.
The behavior log may be used to record user operations on data, each record appearing as: xxx (user name) xxx (data name (such as algorithm name, data set name, etc.)) operation is performed on xxx (sub-module name) of xx (operation button name on right side of data line).
In this embodiment, all users can only view the behavior logs of themselves and lower users. Except for the super administrator, no user can delete any behavior log.
In some embodiments of the present application, the pre-configuration module 480 may be configured to provide independent rights to an administrator, such as the "superadministrator," "enterprise administrator," "project administrator," etc., described above, where the administrator opens independent rights, the iterative training system 400 does not respond to download instructions or update instructions issued by users other than the administrator. That is, this independent right is only available to the administrator, and the "project user" cannot operate this right, and in the case where the administrator opens this right, the "project user" cannot download or update the file.
In some embodiments of the present application, to facilitate invoking the interaction information of the pre-configuration module 480, the pre-configuration module 480 may also be used to provide a path for the visual interface to expose the interaction information, it being understood that the path may be a relative path.
When the user needs to call the corresponding interaction information, the pre-configuration module 480 may be further configured to provide the corresponding interaction information to the user according to the received call instruction, where the call instruction may carry the path of the interaction information. That is, the user needs to use a preset path of the interaction information when invoking the interaction information, otherwise, the user may not operate successfully.
In this embodiment, the pre-configuration module 480 may be used to enable an algorithm engineer to upload an operating environment, an operating script, a description, a usage manual, etc., which is a key for linking the algorithm engineer and an industrial worker.
The pre-configuration module 480 corresponds to a list of files, which may be "running environments" (e.g., python. Exe) or "running scripts" (e.g., test. Py), or may be "specification. Txt", "instruction manual. Pdf", or other conventional files.
In this embodiment, the data list of all pages can support page turning (default 20 pages per page), the text field (except time field) can support fuzzy search, if the table has a time type field, the "time range" filtering can be additionally supported, and all "path" fields can support links on pages (clicking to open the file system).
Referring to fig. 8, fig. 8 is a schematic flow chart of an iterative training method provided in an embodiment of the present application, and the embodiment of the present application provides an iterative training method of an industrial quality inspection model, where the iterative training method may be applied to a client device communicatively connected to a server device. It should be noted that although a logical order is depicted in the flowchart, in some cases the steps shown or described may be performed in a different order than presented. The iterative training method provided by the application specifically comprises the following steps:
step S801, obtaining defect data of a defect product to be inspected through quality inspection equipment arranged in an industrial quality inspection environment, and storing the defect data into a defect product database.
Step S802, providing a visual interface to display defect data in a defect product database, guiding a user to select the defect data to carry out online data marking through visual information, and storing the defect data marked by the user as a model iteration sample into a model iteration data set.
Step 803, the user selects the industrial quality inspection model to be updated, and sends the target industrial quality inspection model selected by the user and the model iteration sample in the model iteration data set to the server-side device, so that the server-side device invokes the training resource to utilize the model iteration sample in the model iteration data set to carry out iteration training on the target industrial quality inspection model.
It should be understood that the steps of the iterative training method provided in the embodiments of the present application may be implemented by respective functional modules included in the iterative training system 400, for example, the steps S801, S802, and S803 may be performed by the data collecting module 410, the data labeling module 420, and the model iteration module 430, respectively. In addition, the iterative training method of the present embodiment may include method steps executed by other functional modules of the iterative training system 400, and for more specific implementation of each step, reference may be made to the above detailed description of each corresponding functional module, which is not repeated herein.
According to the method and the device, the defect data of the defect product are obtained through the quality inspection equipment, the defect data are stored in the defect product database, a visual interface is provided for displaying the defect data, a user is guided to select the defect data to conduct online data marking through visual information, the marked defect data are stored in a model iteration data set as model iteration samples, then the target industrial quality inspection model selected by the user and the model iteration samples in the model iteration data set are sent to the server equipment, the target industrial quality inspection model is subjected to iteration training through the server equipment by utilizing the model iteration samples, the automation degree is high, even industrial workers without computer knowledge can collect the defect data by themselves to iterate the industrial quality inspection model, and the cost is reduced.
To better implement the iterative training system of the present application, as shown in fig. 9, the embodiment of the present application further provides an industrial quality inspection system 900, where the industrial quality inspection system 900 may include a client device 200 and a server device 100 that are communicatively connected;
the client device 200 may be configured to obtain, by using a quality inspection device disposed in an industrial quality inspection environment, defect data of a defect product to be inspected, and store the defect data in a defect product database of the iterative training system; providing a visual interface to display defect data in a defect product database, guiding a user to select the defect data to carry out online data annotation through visual information, and storing the defect data marked by the user as a model iteration sample into a model iteration data set; for the user to select an industrial quality inspection model to be updated, and to send the user-selected target industrial quality inspection model and model iteration samples in the model iteration dataset to the server-side device 100.
The server-side device 100 may be configured to invoke training resources to iteratively train the target industrial quality inspection model using model iteration samples in the model iteration dataset.
The server device 100 may include a plurality of processor resources that may be used to iteratively train the target industrial quality inspection model, for example, may include a plurality of graphics processors (Graphics Processing Unit, GPUs) as shown in fig. 9, and the server device 100 may invoke the GPUs according to the actual situation of the computing resources required for the iterative training. In other embodiments, other types of processor resources may be configured according to the requirements of a specific training model for computing resources, which is not limited in this embodiment.
In the foregoing embodiments, the descriptions of the embodiments are focused on, and the portions of one embodiment that are not described in detail in the foregoing embodiments may be referred to in the foregoing detailed description of other embodiments, which are not described herein again.
In the implementation, each unit or structure may be implemented as an independent entity, or may be implemented as the same entity or several entities in any combination, and the implementation of each unit or structure may be referred to the foregoing embodiments and will not be repeated herein.
The above detailed description of an iterative training system, method and industrial quality testing system for an industrial quality testing model provided in the present application, and specific examples are applied to illustrate the principles and embodiments of the present application, where the above description is only used to help understand the method and core ideas of the present application; meanwhile, as those skilled in the art will vary in the specific embodiments and application scope according to the ideas of the present application, the contents of the present specification should not be construed as limiting the present application in summary.
Claims (10)
1. An iterative training system for an industrial quality inspection model, for application to a client device communicatively coupled to a server device, the iterative training system comprising:
a pre-configuration module for providing interactive information for the iterative training system to enable a user to operate the iterative training system according to the interactive information ；
The data collection module is used for acquiring defect data of a defect product to be inspected through quality inspection equipment arranged in an industrial quality inspection environment and storing the defect data into a defect product database of the iterative training system;
the data marking module is used for providing a visual interface to display the defect data in the defect product database, guiding a user to select the defect data to carry out online data marking through visual information, and storing the defect data marked by the user as a model iteration sample into a model iteration data set;
the model iteration module is used for enabling a user to select an industrial quality inspection model to be updated, and sending a target industrial quality inspection model selected by the user and model iteration samples in the model iteration data set to the server-side equipment so as to call training resources through the server-side equipment to carry out iteration training on the target industrial quality inspection model by using the model iteration samples in the model iteration data set.
2. The iterative training system of claim 1, wherein the defect data is a photograph of a defective product and the quality inspection device is a camera;
the data labeling module is used for:
displaying the photo of the defective product through the visual interface;
displaying marking guide information through the visual interface to guide a user to mark the defect area of the defect product in the displayed photo;
and generating a labeling file comprising labeling information of the defective product based on the labeling result of the photo by the user.
3. The iterative training system of claim 2, wherein the annotation file comprises product label information for the defective product and endpoint information for an annotation box that annotates the defective region.
4. The iterative training system of any of claims 1-3, further comprising:
the model test module is used for a user to select an industrial quality inspection model to be tested, and sends a test sample in a test data set of the model to be tested and the iterative training system selected by the user to the server equipment, so that the server equipment calls a test resource to test the model to be tested by using the test sample in the test data set, and a test record comprising a test result is generated.
5. The iterative training system of claim 1, wherein the interaction information comprises one or more information files in a runtime environment, a runtime script, instructions, and a manual.
6. The iterative training system of claim 1, wherein the pre-configuration module is configured to provide independent rights to an administrator, and wherein the iterative training system is not responsive to download instructions or update instructions issued by a user other than the administrator when the independent rights are opened by the administrator.
7. The iterative training system of claim 1, wherein the pre-configuration module is further configured to provide a visual interface to expose a path of the interaction information.
8. The iterative training system of claim 1, wherein the pre-configuration module is further configured to provide corresponding interaction information to the user according to receiving a call instruction, wherein the call instruction carries a path of the interaction information.
9. An iterative training method for an industrial quality inspection model, applied to a client device communicatively connected to a server device, the iterative training method comprising:
acquiring defect data of a defect product to be inspected through quality inspection equipment arranged in an industrial quality inspection environment, and storing the defect data into a defect product database;
providing a visual interface to display the defect data in the defect product database, guiding a user to select the defect data to carry out online data marking through visual information, and storing the defect data marked by the user as a model iteration sample into a model iteration data set;
and the user selects the industrial quality inspection model to be updated, and sends the target industrial quality inspection model selected by the user and the model iteration sample in the model iteration data set to the server-side equipment so as to call a training resource through the server-side equipment to carry out iterative training on the target industrial quality inspection model by using the model iteration sample in the model iteration data set.
10. An industrial quality inspection system is characterized by comprising a client device and a server device which are in communication connection;
the client device is used for acquiring the defect data of the defect product to be inspected through the quality inspection device arranged in the industrial quality inspection environment and storing the defect data into a defect product database of the iterative training system,
providing a visual interface to display the defect data in the defect product database, guiding a user to select the defect data to carry out online data marking through visual information, storing the defect data marked by the user as a model iteration sample into a model iteration data set,
the method comprises the steps that a user selects an industrial quality inspection model to be updated, and a target industrial quality inspection model selected by the user and a model iteration sample in the model iteration dataset are sent to the server-side equipment;
the server device is used for calling training resources to carry out iterative training on the target industrial quality inspection model by using the model iteration samples in the model iteration data set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311776484.4A CN117763355A (en) | 2023-12-21 | 2023-12-21 | Iterative training system and method for industrial quality inspection model and industrial quality inspection system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311776484.4A CN117763355A (en) | 2023-12-21 | 2023-12-21 | Iterative training system and method for industrial quality inspection model and industrial quality inspection system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117763355A true CN117763355A (en) | 2024-03-26 |
Family
ID=90319608
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311776484.4A Pending CN117763355A (en) | 2023-12-21 | 2023-12-21 | Iterative training system and method for industrial quality inspection model and industrial quality inspection system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117763355A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110554047A (en) * | 2019-09-06 | 2019-12-10 | 腾讯科技（深圳）有限公司 | method, device, system and equipment for processing product defect detection data |
| CN112102263A (en) * | 2020-08-31 | 2020-12-18 | 深圳思谋信息科技有限公司 | Defect detection model generation system, method and device and computer equipment |
| CN112700436A (en) * | 2021-01-13 | 2021-04-23 | 上海微亿智造科技有限公司 | Method, system and medium for improving iteration of industrial quality inspection model |
| CN115496749A (en) * | 2022-11-14 | 2022-12-20 | 江苏智云天工科技有限公司 | Product defect detection method and system based on target detection training preprocessing |
-
2023
- 2023-12-21 CN CN202311776484.4A patent/CN117763355A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110554047A (en) * | 2019-09-06 | 2019-12-10 | 腾讯科技（深圳）有限公司 | method, device, system and equipment for processing product defect detection data |
| CN112102263A (en) * | 2020-08-31 | 2020-12-18 | 深圳思谋信息科技有限公司 | Defect detection model generation system, method and device and computer equipment |
| CN112700436A (en) * | 2021-01-13 | 2021-04-23 | 上海微亿智造科技有限公司 | Method, system and medium for improving iteration of industrial quality inspection model |
| CN115496749A (en) * | 2022-11-14 | 2022-12-20 | 江苏智云天工科技有限公司 | Product defect detection method and system based on target detection training preprocessing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3692438B1 (en) | Automatic generation of a graphic user interface (gui) based on a gui screen image | |
| US11687516B2 (en) | Using an action registry to edit data across users and devices | |
| US11449348B2 (en) | Pre/post deployment customization | |
| US10852898B2 (en) | System for interactive help | |
| CN109783388B (en) | UI (user interface) automatic testing method and device and electronic equipment | |
| US20140082511A1 (en) | Method and system for emulating desktop software applications in a mobile communication network | |
| WO2018036342A1 (en) | Csar-based template design visualization method and device | |
| CA2936397A1 (en) | Method and system for intelligent cloud planning and decommissioning | |
| US11966312B2 (en) | Operation logs visualization device, operation logs visualization method and operation logs visualization program | |
| EP3309695A1 (en) | Information processing apparatus, document display method, document display system, and program | |
| US10915378B1 (en) | Open discovery service | |
| CN116893807A (en) | System and method for designing a robotic flow automation robot using a browser | |
| CN115658496A (en) | Extensible Web automatic test method, system, equipment and storage medium | |
| CN109857462B (en) | Background Docker task mapping method of remote sensing image visual editor | |
| US9063959B2 (en) | Decomposing XML schema documents into subsets | |
| CN105739717B (en) | Data inputting method and device | |
| CN117763355A (en) | Iterative training system and method for industrial quality inspection model and industrial quality inspection system | |
| CN115631374A (en) | Control operation method, control detection model training method, device and equipment | |
| CN109189551A (en) | A kind of method, apparatus and computer storage medium creating middleware | |
| US20220182333A1 (en) | Metadata extraction based on resource attributes and user-defined policy and topology | |
| CN113590564A (en) | Data storage method and device, electronic equipment and storage medium | |
| CN105183490B (en) | Processed offline logic is migrated to the method and apparatus of real-time processing frame | |
| CN113706209B (en) | Operation data processing method and related device | |
| US20240255920A1 (en) | Selective Invocation of RPA Workflows Via API Calls | |
| US20240256370A1 (en) | Selective Invocation of RPA Workflows Via API Calls |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |