CN117475201A - Method, medium and equipment for automatically and iteratively updating model based on network image-text content - Google Patents
Method, medium and equipment for automatically and iteratively updating model based on network image-text content Download PDFInfo
- Publication number
- CN117475201A CN117475201A CN202311353955.0A CN202311353955A CN117475201A CN 117475201 A CN117475201 A CN 117475201A CN 202311353955 A CN202311353955 A CN 202311353955A CN 117475201 A CN117475201 A CN 117475201A
- Authority
- CN
- China
- Prior art keywords
- model
- image
- recognition model
- recognition
- text
- 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
- 238000000034 method Methods 0.000 title claims abstract description 77
- 238000012549 training Methods 0.000 claims abstract description 75
- 238000012795 verification Methods 0.000 claims abstract description 57
- 238000012544 monitoring process Methods 0.000 claims abstract description 13
- 230000004044 response Effects 0.000 claims abstract description 11
- 238000004590 computer program Methods 0.000 claims description 9
- 238000004891 communication Methods 0.000 claims description 2
- 230000000694 effects Effects 0.000 abstract description 15
- 230000008569 process Effects 0.000 description 11
- 238000010586 diagram Methods 0.000 description 10
- 238000012545 processing Methods 0.000 description 5
- 238000013473 artificial intelligence Methods 0.000 description 3
- 230000007423 decrease Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000007781 pre-processing Methods 0.000 description 3
- 230000003190 augmentative effect Effects 0.000 description 2
- 238000005530 etching Methods 0.000 description 2
- 238000013210 evaluation model Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000011084 recovery Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 238000003491 array Methods 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/776—Validation; Performance evaluation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- Image Analysis (AREA)
Abstract
The application provides a method, medium and equipment for automatically and iteratively updating a model based on network image-text content. The method comprises the following steps: in the online operation of the image-text recognition model, acquiring missed recognition error recognition data of the image-text recognition model, and adding the missed recognition error recognition data into a model training sample library; monitoring the recognition degree of the image-text recognition model; in response to the recognition degree of the image-text recognition model being lower than a recognition degree threshold, training the image-text recognition model based on the model training sample library to generate a new image-text recognition model; performing on-line gray level verification on the new image-text recognition model; and switching the image-text recognition model to online operation of the new image-text recognition model in response to the verification result of the online gray level verification meeting a preset value. The method and the device can automatically classify and collect training samples, automatically monitor the effect of the image-text recognition model, and realize the automatic training and iteration of the image-text recognition model with safe content.
Description
Technical Field
The application belongs to the technical field of artificial intelligence, and particularly relates to the technical field of model training of artificial intelligence.
Background
Image recognition technology and text recognition technology are becoming more and more widespread in internet services, and more companies make services increasingly efficient through these technology capabilities. Such as community type business, uses image recognition and text recognition model to automatically process illegal and illegal data. However, because the model training is defective and the network environment is continuously changed, the accuracy and recall rate of the model often decrease due to the generation of new content in the period of time, and the model effect needs to be maintained by continuing the iterative update.
In order to improve the accuracy of the model algorithm in the service and reduce the missing killing and the miskilling, the model is generally retrained by using a newly generated sample after a period of time, but the model is retrained manually in the prior art, the training process is repeated, and the human resources are consumed.
Disclosure of Invention
The application provides a method, medium and equipment for automatically iterating and updating a model based on network image-text content, which are used for automatically training and iterating image-text models and solving the problem of model effect decline or insufficient coverage.
In a first aspect, an embodiment of the present application provides a method for automatically iteratively updating a model based on web content, including: in the online operation of the image-text recognition model, acquiring missed recognition error recognition data of the image-text recognition model, and adding the missed recognition error recognition data into a model training sample library; monitoring the recognition degree of the image-text recognition model; in response to the recognition degree of the image-text recognition model being lower than a recognition degree threshold, training the image-text recognition model based on the model training sample library to generate a new image-text recognition model; performing on-line gray level verification on the new image-text recognition model; and switching the image-text recognition model to online operation of the new image-text recognition model in response to the verification result of the online gray level verification meeting a preset value.
In an implementation manner of the first aspect, the method further includes: and establishing a regular standard-reaching task for the model training sample library so as to classify the missing identification data in the model training sample library according to standard.
In an implementation manner of the first aspect, the monitoring the recognition degree of the image-text recognition model includes: acquiring log data of the image-text recognition model and the manual auditing platform; and calculating the recognition degree of the image-text recognition model based on the log data.
In an implementation manner of the first aspect, the monitoring the recognition degree of the image-text recognition model includes: obtaining auditing data from a manual auditing platform, and obtaining the recognition result of the image-text recognition model; and calculating the recognition degree of the image-text recognition model based on the auditing data and the recognition result.
In an implementation manner of the first aspect, the recognition degree includes a recall rate and an accuracy rate of the image-text recognition model in a corresponding application scenario.
In an implementation manner of the first aspect, the method further includes: acquiring a manual auditing and verifying sample set from the model training sample library; and carrying out on-line verification on the new image-text type recognition model based on the manual verification sample set.
In an implementation manner of the first aspect, the performing on-line gray scale verification on the new image-text recognition model includes: reading a new image-text recognition model, and establishing a plurality of model services for the new image-text recognition model; and acquiring subscription user data of the application scene, and acquiring a verification result of a new image-text type recognition model based on the subscription user data of the application scene and a plurality of model services.
In an implementation manner of the first aspect, the method further includes: and acquiring user configuration information, and subscribing the subscription user data of the application scene according to the user configuration information.
In a second aspect, embodiments of the present application provide a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements a method for automatically iteratively updating a model based on web content according to any of the first aspects of the present application.
In a third aspect, an embodiment of the present application provides an electronic device, including: a memory storing a computer program; and the processor is in communication connection with the memory, and executes the method for automatically and iteratively updating the model based on the network image-text content according to any one of the first aspect of the application when the computer program is called.
The method for automatically iteratively updating the model based on the network image-text content can automatically classify and collect training samples and automatically monitor the effect of the model, and achieves automatic training and iteration of the content security model.
Drawings
Fig. 1 is a schematic application view of a method for automatically and iteratively updating a model based on web content according to an embodiment of the present application.
Fig. 2 is a flowchart of a method for automatically iteratively updating a model based on web content according to an embodiment of the present application.
FIG. 3 is a flow chart of one implementation of a method for automatically iteratively updating a model based on web content according to an embodiment of the present application.
FIG. 4 is a flow chart illustrating model training in a method for automatically iteratively updating models based on web content according to an embodiment of the present application.
Fig. 5 shows a flowchart of online verification set verification in a method for automatically and iteratively updating a model based on web content according to an embodiment of the present application.
Fig. 6 is a flowchart of performing online gray scale verification on a new image-text recognition model in a method for automatically and iteratively updating models based on web image-text content according to an embodiment of the present application.
Fig. 7 is a schematic diagram of an implementation principle of performing online gray scale verification on a new image-text recognition model in a method for automatically and iteratively updating models based on web image-text content according to an embodiment of the present application.
Fig. 8 is a schematic diagram of an implementation process of performing online gray scale verification on a new image-text recognition model in a method for automatically and iteratively updating models based on web image-text content according to an embodiment of the present application.
Fig. 9 is a schematic diagram of an overall implementation process of a method for automatically and iteratively updating a model based on web content according to an embodiment of the present application.
Fig. 10 is a schematic diagram of an application example of a method for automatically and iteratively updating a model based on web content according to an embodiment of the present application.
Fig. 11 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
Description of element reference numerals
100. Electronic equipment
101. Memory device
102. Processor and method for controlling the same
103. Display device
S100 to S500 steps
S210 to S220 steps
S410 to S420 steps
Detailed Description
Other advantages and effects of the present application will become apparent to those skilled in the art from the present disclosure, when the following description of the embodiments is taken in conjunction with the accompanying drawings. The present application may be embodied or carried out in other specific embodiments, and the details of the present application may be modified or changed from various points of view and applications without departing from the spirit of the present application. It should be noted that the following embodiments and features in the embodiments may be combined with each other without conflict.
The embodiment of the application provides a method for automatically and iteratively updating a model based on network image-text content, which is used for automatically training and iterating image-text models and solving the problem of model effect decline or insufficient coverage.
Fig. 1 is a schematic application view of a method for automatically and iteratively updating a model based on web content according to an embodiment of the present application. As shown in fig. 1, the method for automatically and iteratively updating a model based on network graphic content according to the present embodiment is applicable to a graphic recognition model, where the graphic recognition model includes an image recognition model, a text recognition model, or a graphic recognition model capable of recognizing an image and a text at the same time. After the image-text recognition model in the embodiment is in online operation, data of missing recognition or wrong recognition of the image-text recognition model is collected and automatically gathered into a model training sample library, so that automatic classification and training sample collection are realized. And meanwhile, when the image-text recognition model runs on line, monitoring the effect (recognition degree) of the image-text recognition model to realize an automatic monitoring model, if the model effect is reduced to a set water line, triggering an automatic model training task, and retraining the image-text recognition model according to training sample data updated in a model training sample library. And then, carrying out online gray level verification on the retrained image-text recognition model, and if the verification result meets the requirement, switching the image-text recognition model which is currently operated online into a new image-text recognition model to operate online, namely replacing the original model with the new version model to be applied to online service, so as to realize automatic training and iteration of the model.
The following describes the technical solution in the embodiment of the present application in detail with reference to fig. 2 to 10 in the embodiment of the present application.
The embodiment provides a method for automatically and iteratively updating a model based on network image-text content, and fig. 2 is a flowchart of the method for automatically and iteratively updating the model based on network image-text content in the embodiment of the application. As shown in fig. 2, the method for automatically and iteratively updating a model based on web content according to the embodiment of the present application includes the following steps S100 to S400.
Step S100, collecting missed recognition data of an image-text recognition model in online operation of the image-text recognition model, and adding the missed recognition data into a model training sample library;
step S200, monitoring the recognition degree of the image-text recognition model;
step S300, training the image-text recognition model based on the model training sample library to generate a new image-text recognition model in response to the recognition degree of the image-text recognition model being lower than a recognition degree threshold;
step S400, performing on-line gray verification on the new image-text recognition model;
step S500: and switching the image-text recognition model to online operation of the new image-text recognition model in response to the verification result of the online gray level verification meeting a preset value.
The following describes the steps S100 to S500 of the method for automatically and iteratively updating a model based on the web content in this embodiment in detail.
And step S100, collecting missed recognition data of the image-text recognition model in the online operation of the image-text recognition model, and adding the missed recognition data into a model training sample library.
In this embodiment, the method for collecting the missed recognition data of the image-text recognition model includes a method for obtaining the missed recognition data of the image-text recognition model through manual auditing, or a method for detecting the recognition result of the image-text recognition model through a machine model to obtain the missed recognition data of the image-text recognition model.
In this embodiment, the missed recognition error recognition data may be added to the model training sample library by means of message subscription.
In one implementation of this embodiment, the method further includes: and establishing a regular standard-reaching task for the model training sample library so as to classify the missing identification data in the model training sample library according to standard. And performing marking classification manually or performing standard classification on missed recognition data in the model training sample library through a pre-trained machine classification model.
Therefore, after the image-text recognition model is in online operation, the data of missing recognition or wrong recognition of the image-text recognition model is collected, and the data is automatically collected into a model training sample library in a message throwing mode, so that automatic classification and training sample collection are realized. Therefore, the embodiment realizes automatic collection of training samples according to the labels, and can mark and fall down in real time, thereby saving the cost of collecting and processing the training samples by a model team.
Step S200, monitoring the recognition degree of the image-text recognition model.
In one implementation manner of this embodiment, the recognition degree includes recall rate and accuracy rate of the image-text recognition model in a corresponding application scenario.
The recall rate and the accuracy rate are calculated as follows:
accuracy: accuracy= (tp+tn)/(tp+tn+fp+fn);
recall rate recovery: recovery = TP/(TP/FN);
wherein TP is true positive, TN is true negative, FP is false positive, and FN is false negative.
FIG. 3 is a flow chart of one implementation of a method for automatically iteratively updating a model based on web content according to an embodiment of the present application. As shown in fig. 3, in one implementation manner of this embodiment, the manner of monitoring the recognition degree of the image-text recognition model includes:
step S210, acquiring the image-text recognition model and log data of a manual auditing platform;
step S220, calculating the recognition degree of the image-text recognition model based on the log data.
For example, the big data computing platform collects and computes the detection logs of the auditing platform and the model platform to obtain identification results such as true positives, true negatives, false positives or false negatives, and then calculates the recall rate and the accuracy rate of each model under each scene.
In addition, in an implementation manner of this embodiment, the method for monitoring the recognition degree of the image-text recognition model may also include:
1) Obtaining auditing data from a manual auditing platform, and obtaining the recognition result of the image-text recognition model;
2) And calculating the recognition degree of the image-text recognition model based on the auditing data and the recognition result.
For example, after the auditing platform performs manual auditing, each piece of data is thrown to the model platform, the model platform records the model result and the manual auditing state, and the identification results such as true positives, true negatives, false positives, or false negatives and the manual auditing state are obtained, so that the recall rate and the accuracy rate of each model under each scene are calculated at regular time.
And step S300, training the image-text recognition model based on the model training sample library to generate a new image-text recognition model in response to the recognition degree of the image-text recognition model being lower than a recognition degree threshold.
FIG. 4 is a flow chart illustrating model training in a method for automatically iteratively updating models based on web content according to an embodiment of the present application. As shown in fig. 4, training the image-text recognition model based on the model training sample library, and generating a new image-text recognition model includes three stages: a preprocessing stage, a training stage and a verification stage. The preprocessing stage obtains model training parameters, initializes the model, and then obtains a model training sample set from a model training sample library, wherein new training samples are updated in the model training sample set. And then entering a model training stage, training the image-text type recognition model which is currently operated online according to the updated training sample, generating a new image-text type recognition model, and then storing a new model file. After training the new model, entering a verification stage, verifying the evaluation model according to a preset online verification set, generating a state notification to prompt if verification is passed, and creating a model version.
Fig. 5 shows a flowchart of online verification set verification in a method for automatically and iteratively updating a model based on web content according to an embodiment of the present application. In one implementation of this embodiment, the method includes: acquiring a manual auditing and verifying sample set from the model training sample library; and carrying out on-line verification on the new image-text type recognition model based on the manual verification sample set.
Specifically, as shown in fig. 5, the online verification set verification flow includes:
and loading a new version model file, then acquiring an on-line manually judged sample (namely manually audited data) from a sample data platform, carrying out reasoning and identification through the new image-text identification model which is newly trained, and calculating and storing an identification result. If the new image-text recognition model passes the verification of the online verification set, a state notification is generated to prompt, and a model version is newly built. According to the embodiment, the link of training the repetitive model by the model team is changed into an automatic process to be realized, and the iteration efficiency is greatly improved. By adopting the method of online assembling multiple mills, the method can avoid integrally publishing multiple models again, support independent publishing of sub-models, improve iteration efficiency and avoid data loss in the version switching process.
Step S400, performing on-line gray level verification on the new image-text recognition model.
After training the new image-text recognition model, verifying the online flow gray scale effect of the new version model, namely verifying the recall rate and the accuracy rate of the new image-text recognition model.
The gray level verification, namely gray level test, is a test method for the model before the model is on line.
Fig. 6 is a flowchart of performing online gray scale verification on a new image-text recognition model in a method for automatically and iteratively updating models based on web image-text content according to an embodiment of the present application. As shown in fig. 6, in an implementation manner of this embodiment, the performing on-line gray scale verification on the new image-text recognition model includes:
step S410, reading a new image-text recognition model, and establishing a plurality of model services for the new image-text recognition model;
step S420, acquiring subscription user data of an application scene, and acquiring a verification result of a new image-text type recognition model based on the subscription user data of the application scene and a plurality of model services.
Wherein, in an implementation manner of this embodiment, further includes: and acquiring user configuration information, and subscribing the subscription user data of the application scene according to the user configuration information.
Fig. 7 is a schematic diagram of an implementation principle of performing online gray scale verification on a new image-text recognition model in a method for automatically and iteratively updating models based on web image-text content according to an embodiment of the present application. As shown in fig. 7, a model version is obtained from a model file repository, a plurality of model services are established for a new image-text type recognition model, subscription user data of an application scene is obtained from a message subscription tool such as kafka, and verification results of the new image-text type recognition model are obtained based on the subscription user data of the application scene and the plurality of model services.
Fig. 8 is a schematic diagram of an implementation process of performing online gray scale verification on a new image-text recognition model in a method for automatically and iteratively updating models based on web image-text content according to an embodiment of the present application. As shown in fig. 8, when performing the on-line gray scale verification process, the new version model file is read, the model service is automatically created, the on-line re-etching flow message of the specific scene is subscribed according to the user configuration information, and then the subscribed user data of the application scene is processed by the plurality of model services, so as to obtain the verification result of the new image-text type identification model.
The method for assembling multiple models on line is adopted, so that the multiple models can be prevented from being released again integrally, independent release of the submodels is supported, iteration efficiency is improved, and data loss in the version switching process is avoided.
And step S500, switching the image-text recognition model to the online operation of the new image-text recognition model in response to the verification result of the online gray level verification meeting a preset value.
If the recall rate and the accuracy rate effect are met, the new image-text identification model is switched into the online flow, and the online flow is applied to production.
Fig. 9 is a schematic diagram of an overall implementation process of a method for automatically and iteratively updating a model based on web content according to an embodiment of the present application. As shown in fig. 9, after the image-text recognition model is in online operation, data of missing recognition or misrecognition of the image-text recognition model is collected, and the data are automatically collected into a model training sample library in a message throwing mode, so that automatic classification and training sample collection are realized. Therefore, the embodiment realizes automatic collection of training samples according to the labels, and can mark and fall down in real time, thereby saving the cost of collecting and processing the training samples by a model team. The embodiment can calculate the recognition degree of the image-text recognition model based on the image-text recognition model and the log data of the manual auditing platform, and can calculate the recognition degree of the image-text recognition model based on the auditing data acquired from the manual auditing platform and the recognition result. If the model effect is reduced to the set water line, triggering an automatic model training task, training the image-text type recognition model of the current online operation according to the updated training sample, generating a new image-text type recognition model, and then storing a new model file. After training the new model, entering a verification stage, verifying the evaluation model according to a preset online verification set, generating a state notification to prompt if verification is passed, and creating a model version. After training the new image-text recognition model, carrying out on-line flow gray scale effect verification on the new version model, namely verifying the recall rate and the accuracy rate of the new image-text recognition model. And when the online gray level verification process is carried out, reading the new version model file, automatically creating a model service, subscribing the online re-etching flow information of the specific scene according to the user configuration information, and then processing the subscribed user data of the application scene by a plurality of model services to obtain a verification result of the new image-text identification model. If the recall rate and the accuracy rate effect are met, the new image-text identification model is switched into the online flow, and the online flow is applied to production.
Fig. 10 is a schematic diagram of an application example of a method for automatically and iteratively updating a model based on web content according to an embodiment of the present application. As shown in fig. 10, in this embodiment, by opening the auditing system, the model trains the sample data platform and the model platform, and realizes automatic training and iteration of the content security model; through channels such as manual judgment of a service system, training samples are automatically classified and collected, preprocessing such as marking data at any time is supported, the model can automatically acquire sample data according to types, and automatic training is performed according to preset conditions; the on-line model effect and the new version model can be predicted according to the type.
The protection scope of the method for automatically updating the model based on the network graphics context is not limited to the execution sequence of the steps listed in the embodiment, and all the schemes implemented by adding or removing steps and replacing steps according to the principles of the present application in the prior art are included in the protection scope of the present application.
The embodiment of the application also provides a computer readable storage medium, on which a computer program is stored, which when executed by a processor, implements the method for automatically and iteratively updating a model based on web content provided by any embodiment of the application.
Any combination of one or more storage media may be employed in embodiments of the present application. The storage medium may be a computer readable signal medium or a computer readable storage medium. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a RAM, a ROM, an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
The embodiment of the application also provides electronic equipment. Fig. 11 is a schematic structural diagram of an electronic device 100 according to an embodiment of the present application. In some embodiments, the electronic device may be a mobile phone, tablet, wearable device, in-vehicle device, augmented Reality (Augmented Reality, AR)/Virtual Reality (VR) device, notebook, ultra-Mobile Personal Computer (UMPC), netbook, personal digital assistant (Personal Digital Assistant, PDA), or other terminal device. In addition, the method based on the automatic iterative updating model of the network image-text content can be applied to databases, servers and service response systems based on terminal artificial intelligence. The embodiment of the application does not limit the specific application scene of the method for automatically and iteratively updating the model based on the network image-text content.
As shown in fig. 11, an electronic device 100 provided in an embodiment of the present application includes a memory 101 and a processor 102.
The memory 101 is for storing a computer program; preferably, the memory 101 includes: various media capable of storing program codes, such as ROM, RAM, magnetic disk, U-disk, memory card, or optical disk.
In particular, memory 101 may include computer system readable media in the form of volatile memory, such as Random Access Memory (RAM) and/or cache memory. Electronic device 100 may further include other removable/non-removable, volatile/nonvolatile computer system storage media. Memory 101 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the various embodiments of the present application.
The processor 102 is connected to the memory 101 and is configured to execute a computer program stored in the memory 101, so that the electronic device 100 performs the method for automatically and iteratively updating a model based on web content provided in any embodiment of the present application.
Alternatively, the processor 102 may be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU), a network processor (Network Processor, NP), etc.; but also digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), field programmable gate arrays (Field Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
Optionally, the electronic device 100 in this embodiment may further include a display 103. A display 103 is communicatively coupled to the memory 101 and the processor 102 for displaying a related GUI interactive interface for a method for automatically iteratively updating a model based on web content.
In summary, the method for automatically iteratively updating the model based on the network image-text content provided by the embodiment of the application can automatically classify and collect training samples and automatically monitor the effect of the model, thereby realizing the automatic training and iteration of the content security model. Therefore, the method effectively overcomes various defects in the prior art and has high industrial utilization value.
The foregoing embodiments are merely illustrative of the principles of the present application and their effectiveness, and are not intended to limit the application. Modifications and variations may be made to the above-described embodiments by those of ordinary skill in the art without departing from the spirit and scope of the present application. Accordingly, it is intended that all equivalent modifications and variations which may be accomplished by persons skilled in the art without departing from the spirit and technical spirit of the disclosure be covered by the claims of this application.
Claims (10)
1. A method for automatically iteratively updating a model based on web content, comprising:
in the online operation of the image-text recognition model, acquiring missed recognition error recognition data of the image-text recognition model, and adding the missed recognition error recognition data into a model training sample library;
monitoring the recognition degree of the image-text recognition model;
in response to the recognition degree of the image-text recognition model being lower than a recognition degree threshold, training the image-text recognition model based on the model training sample library to generate a new image-text recognition model;
performing on-line gray level verification on the new image-text recognition model;
and switching the image-text recognition model to online operation of the new image-text recognition model in response to the verification result of the online gray level verification meeting a preset value.
2. The method for automatically iteratively updating models based on web content of claim 1, further comprising:
and establishing a regular standard-reaching task for the model training sample library so as to classify the missing identification data in the model training sample library according to standard.
3. The method for automatically iteratively updating models based on web content according to claim 1, wherein the monitoring the recognition of the graphic recognition model comprises:
acquiring log data of the image-text recognition model and the manual auditing platform;
and calculating the recognition degree of the image-text recognition model based on the log data.
4. The method for automatically iteratively updating models based on web content according to claim 1, wherein the monitoring the recognition of the graphic recognition model comprises:
obtaining auditing data from a manual auditing platform, and obtaining the recognition result of the image-text recognition model;
and calculating the recognition degree of the image-text recognition model based on the auditing data and the recognition result.
5. The method for automatically and iteratively updating models based on network teletext content according to claim 3 or 4, wherein the recognition degree includes recall rate and accuracy rate of the teletext class recognition model in a corresponding application scenario.
6. The method for automatically iteratively updating models based on web content of claim 1, further comprising:
acquiring a manual auditing and verifying sample set from the model training sample library;
and carrying out on-line verification on the new image-text type recognition model based on the manual verification sample set.
7. The method for automatically iteratively updating models based on web content according to claim 1, wherein the performing on-line gray scale verification on the new graphic class identification model comprises:
reading a new image-text recognition model, and establishing a plurality of model services for the new image-text recognition model;
and acquiring subscription user data of the application scene, and acquiring a verification result of a new image-text type recognition model based on the subscription user data of the application scene and a plurality of model services.
8. The method for automatically iteratively updating models based on web content of claim 7, further comprising:
and acquiring user configuration information, and subscribing the subscription user data of the application scene according to the user configuration information.
9. A computer readable storage medium having stored thereon a computer program, which when executed by a processor implements the method of automatically iteratively updating a model based on web content of any of claims 1 to 8.
10. An electronic device, the electronic device comprising:
a memory storing a computer program;
a processor, in communication with the memory, for executing the method of automatically iteratively updating a model based on web content of any of claims 1 to 8 when the computer program is invoked.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311353955.0A CN117475201A (en) | 2023-10-18 | 2023-10-18 | Method, medium and equipment for automatically and iteratively updating model based on network image-text content |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311353955.0A CN117475201A (en) | 2023-10-18 | 2023-10-18 | Method, medium and equipment for automatically and iteratively updating model based on network image-text content |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117475201A true CN117475201A (en) | 2024-01-30 |
Family
ID=89632280
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311353955.0A Pending CN117475201A (en) | 2023-10-18 | 2023-10-18 | Method, medium and equipment for automatically and iteratively updating model based on network image-text content |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117475201A (en) |
-
2023
- 2023-10-18 CN CN202311353955.0A patent/CN117475201A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111522704A (en) | Alarm information processing method, device, computer device and storage medium | |
| CN112148772A (en) | Alarm root cause identification method, device, equipment and storage medium | |
| CN109992484B (en) | Network alarm correlation analysis method, device and medium | |
| CN110928862A (en) | Data cleaning method, data cleaning apparatus, and computer storage medium | |
| CN112434178B (en) | Image classification method, device, electronic equipment and storage medium | |
| CN112883990A (en) | Data classification method and device, computer storage medium and electronic equipment | |
| CN113420673B (en) | Garbage classification method, device, equipment and storage medium | |
| CN116414948A (en) | Abnormal data mining method and software product based on cloud data and artificial intelligence | |
| CN113780367A (en) | Classification model training and data classification method and device, and electronic equipment | |
| CN113704389A (en) | Data evaluation method and device, computer equipment and storage medium | |
| CN113111804B (en) | Face detection method and device, electronic equipment and storage medium | |
| CN115984158A (en) | Defect analysis method and device, electronic equipment and computer readable storage medium | |
| CN112579781B (en) | Text classification method, device, electronic equipment and medium | |
| CN110782128B (en) | User occupation label generation method and device and electronic equipment | |
| CN111178037A (en) | Repeated defect report identification method and device and electronic equipment | |
| CN117475201A (en) | Method, medium and equipment for automatically and iteratively updating model based on network image-text content | |
| CN113590484B (en) | Algorithm model service testing method, system, equipment and storage medium | |
| CN114691505A (en) | Method for locating program problem, electronic device and storage medium | |
| CN112465149A (en) | Same-city part identification method and device, electronic equipment and storage medium | |
| CN113850686A (en) | Insurance application probability determination method and device, storage medium and electronic equipment | |
| CN113469816A (en) | Digital currency identification method, system and storage medium based on multigroup technology | |
| CN113535458A (en) | Abnormal false alarm processing method and device, storage medium and terminal | |
| CN113743293A (en) | Fall behavior detection method and device, electronic equipment and storage medium | |
| CN113344064A (en) | Event processing method and device | |
| CN118197027B (en) | Unmanned aerial vehicle scheduling checking method based on pipeline early warning and related device |
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 |