Method, device and storage medium for dynamically updating data sets

Abstract

The present application discloses a method, device and storage medium for dynamically updating a data set, which is used to improve the data quality and model performance of model training. The method includes obtaining a training data set for a detection model, where the detection model is an artificial intelligence model that will be added to the current display screen production line to perform display screen defect detection; obtaining a display screen acquisition data set in real time, where the display screen acquisition data set is data collected in the current display screen production line for display screen defect detection; performing a data freshness assessment on the display screen acquisition data set; performing a data importance assessment on the display screen acquisition data set; performing an update level analysis on the display screen acquisition data set based on the data freshness assessment results and the data importance assessment results; filtering out a target data item set from the display screen acquisition data set based on the update level analysis results; and updating the filtered target data item set into the training data set.

Description

Technical Field

[0001] The present application relates to the field of display screen detection, and in particular to a method, device, and storage medium for dynamically updating a data set. Background Art

[0002] In the field of display screen inspection, it is often necessary to detect different types of defects on the display screen. Defect detection and defect compensation are important production processes. When different display screen structures generate the same defect, its manifestation is also different. Therefore, on the display screen production line, multiple artificial intelligence models are usually used to detect different display defects, such as detecting defects in the display screen body, detecting defects in the display screen, detecting defects in the flat part of the display screen, detecting defects in the curved part of the display screen, and so on. However, with the continuous update and iteration of display screens, the functions on the display screens have gradually increased and become increasingly refined, which has also brought more complex changes to the display screen structure. For example, to increase functionality, a new thin-film circuit structure is added below the pixel layer in the display screen. This circuit structure has a certain degree of reflectivity. If there is a defect in the pixel layer area above this circuit area, it may cause false detection or missed detection. To cope with the defect detection of new display screens, it is often chosen to add new artificial intelligence defect detection models to the display screen production line. While completing targeted defect detection, it also reduces the cost of production line changes.

[0003] After the inspection model is designed, it is trained using a training dataset (historical display screen data) before deployment. However, deployment is often not possible directly at this point because the training data used by the inspection model is often relatively old, while new display screens (with new structures) vary from batch to batch. Therefore, fresher data is often required for training before the inspection model is deployed. Data freshness has a significant impact on the performance of the inspection model after deployment. However, existing training dataset update methods are mostly static, requiring manual selection of the latest batch of display screen data. Furthermore, because each display screen ID is subject to multiple defect detection images for the same inspection project, the amount of display screen data collected is often very large, requiring additional screening of more important display screen data. This is time-consuming and inefficient, and it prevents staff from accurately selecting display screen data from the current production line that best matches the inspection model, i.e., timely reflecting the latest changes in display screen data. This results in poor pre-deployment training results, reducing training data quality and inspection model performance. Summary of the Invention

[0004] The present application discloses a method, device and storage medium for dynamically updating a data set, which are used to improve data quality and model performance of model training.

[0005] In a first aspect, an embodiment of the present application provides a method for dynamically updating a data set, including: obtaining a training data set of a detection model, where the detection model is an artificial intelligence model that will be added to the current display screen production line to perform display screen defect detection, and the training data set is the display screen data used by the detection model during pre-training; obtaining a display screen acquisition data set in real time, where the display screen acquisition data set is data collected in the current display screen production line for display screen defect detection; performing a data freshness evaluation on the display screen acquisition data set; performing a data importance evaluation on the display screen acquisition data set; performing an update level analysis on the display screen acquisition data set based on the data freshness evaluation results and the data importance evaluation results; filtering out a target data item set from the display screen acquisition data set based on the update level analysis results; and updating the filtered target data item set into the training data set.

[0006] Optionally, after the step of updating the screened target data item set into the training data set, the method further includes: using the updated training data set to pre-train the detection model before it is put into use; evaluating the model performance of the detection model using the updated training data set to generate performance feedback information; when the performance feedback information shows that the detection model performance meets the standards, deploying the detection model to the display production line; and sending the performance feedback information of the detection model back to the production client for optimization in future model training.

[0007] Optionally, after the step of updating the screened target data item set into the training data set, and before the step of pre-training the detection model before using the updated training data set, the method further includes: pre-training sorting the display screen collection data in the target data item set according to the update level analysis results.

[0008] Optionally, the detection model is an artificial intelligence model for defect detection of the circuit area of the display screen; after the step of obtaining the display screen acquisition data set in real time and before the step of evaluating the data freshness of the display screen acquisition data set, the method also includes: filtering out the display screen acquisition data containing the circuit area from the display screen acquisition data set, and obtaining the generation time data of the display screen acquisition data containing the circuit area; performing defect area and defect type analysis on the display screen acquisition data containing the circuit area, and generating a defect feature label for each display screen acquisition data with a defect in the circuit area; classifying the display screen acquisition data after the defect area and defect type analysis according to the display screen ID and the detection item; generating a circuit structure hierarchy based on the classified display screen acquisition data, the circuit structure hierarchy representing the structural level of the thin-film circuit in the circuit area; binding the obtained generation time data, defect feature label, and circuit structure hierarchy with the classified display screen acquisition data to generate a new display screen acquisition data set.

[0009] Optionally, the step of performing data freshness evaluation on the display screen collection data set includes: selecting a target display screen collection data from the new display screen collection data set; determining a time decay function based on the defect detection items of the target display screen collection data, where the time decay function has an unknown variable, namely, a generation time; generating a business weight based on the display screen model of the target display screen collection data, where the business weight refers to the importance of the display screen corresponding to the display screen model to the training of the detection model; determining a data change rate parameter corresponding to the display screen model based on the display screen model of the target display screen collection data and the new display screen collection data set; performing freshness evaluation based on the time decay function, the business weight, the data change rate parameter, and the generation time of the target display screen collection data; performing freshness evaluation on other display screen collection data in the new display screen collection data set, and integrating them to form a data freshness evaluation result.

[0010] Optionally, the step of performing data importance evaluation on the display screen acquisition data set includes: generating a structural weight according to a circuit structure hierarchy of the target display screen acquisition data; determining a target display screen ID of the target display screen acquisition data, and analyzing a binding status of defect feature labels of all display screen acquisition data of the same inspection project under the target display screen ID; generating a detection weight for the target display screen acquisition data according to the binding analysis result; analyzing the defect feature labels of all display screen acquisition data of the same inspection project under the target display screen ID, and generating a defect weight for the target display screen acquisition data according to a type analysis result of the defect feature label; performing importance evaluation on the target display screen acquisition data according to the structural weight, the detection weight, and the defect weight; performing importance evaluation on other display screen acquisition data in the new display screen acquisition data set, and integrating them to form a data importance evaluation result.

[0011] Optionally, the step of performing an update level analysis on the display screen collected data set according to the data freshness evaluation result and the data importance evaluation result includes: obtaining a first freshness threshold, a second freshness threshold, a first importance score threshold and a second importance score threshold, the first freshness threshold being less than the second freshness threshold, and the first importance score threshold being less than the second importance score threshold; generating a freshness level for each display screen collected data in the display screen collected data set according to the data freshness evaluation result, the first freshness threshold and the second freshness threshold; generating an importance level for each display screen collected data in the display screen collected data set according to the data importance evaluation result, the first importance score threshold and the second importance score threshold; performing an update level analysis on each display screen collected data in the display screen collected data set according to the freshness level and the importance level, and generating an update level analysis result.

[0012] In a second aspect, an embodiment of the present application provides a device for dynamically updating a data set, comprising: a first acquisition unit for acquiring a training data set of a detection model, where the detection model is an artificial intelligence model to be added to the current display screen production line to perform display screen defect detection, and the training data set is the display screen data used by the detection model during pre-training; a second acquisition unit for acquiring a display screen acquisition data set in real time, where the display screen acquisition data set is data collected in the current display screen production line for display screen defect detection; a first evaluation unit for performing data freshness evaluation on the display screen acquisition data set; a second evaluation unit for performing data importance evaluation on the display screen acquisition data set; a first analysis unit for performing an update level analysis on the display screen acquisition data set based on the data freshness evaluation results and the data importance evaluation results; a screening unit for screening out a target data item set from the display screen acquisition data set based on the update level analysis results; and an update unit for updating the screened target data item set into the training data set.

[0013] Optionally, after the updating unit, the device further includes: a pre-training unit, used to use the updated training data set to pre-train the detection model before it is put into use; a first generation unit, used to evaluate the model performance of the detection model using the updated training data set and generate performance feedback information; a deployment unit, used to deploy the detection model to the display production line when the performance feedback information shows that the detection model performance meets the standards; and a feedback unit, used to send the performance feedback information of the detection model back to the production client for optimization in future model training.

[0014] Optionally, after the updating unit and before the pre-training unit, the apparatus further includes: a sorting unit configured to perform pre-training sorting on the display screen collected data in the target data item set according to the update level analysis result.

[0015] Optionally, the detection model is an artificial intelligence model for defect detection of the circuit area of the display screen; after the second acquisition unit and before the first evaluation unit, the device also includes: a third acquisition unit, which is used to filter out the display screen acquisition data containing the circuit area from the display screen acquisition data set, and obtain the generation time data of the display screen acquisition data containing the circuit area; a second analysis unit, which is used to perform defect area and defect type analysis on the display screen acquisition data containing the circuit area, and generate a defect feature label for each display screen acquisition data with a defect in the circuit area; a classification unit, which is used to classify the display screen acquisition data after the defect area and defect type analysis according to the display screen ID and the detection item; a second generation unit, which is used to generate a circuit structure hierarchy based on the classified display screen acquisition data, and the circuit structure hierarchy represents the structural level of the thin-film circuit in the circuit area; a third generation unit, which is used to bind the acquired generation time data, defect feature label, and circuit structure hierarchy with the classified display screen acquisition data to generate a new display screen acquisition data set.

[0016] Optionally, the steps of the first evaluation unit include: selecting a target display screen acquisition data from the new display screen acquisition data set; determining a time decay function based on the defect detection items of the target display screen acquisition data, where the time decay function has an unknown variable, namely, generation time; generating a business weight based on the display screen model of the target display screen acquisition data, where the business weight refers to the importance of the display screen corresponding to the display screen model to the training of the detection model; determining a data change rate parameter corresponding to the display screen model based on the display screen model of the target display screen acquisition data and the new display screen acquisition data set; performing a freshness evaluation based on the time decay function, the business weight, the data change rate parameter, and the generation time of the target display screen acquisition data; performing a freshness evaluation on other display screen acquisition data in the new display screen acquisition data set, and integrating them to form a data freshness evaluation result.

[0017] Optionally, the steps of the second evaluation unit include: generating a structural weight according to a circuit structure hierarchy of the target display screen acquisition data; determining a target display screen ID of the target display screen acquisition data, and analyzing a binding status of defect feature labels of all display screen acquisition data of the same detection item under the target display screen ID; generating a detection weight for the target display screen acquisition data according to the binding analysis result; analyzing the defect feature labels of all display screen acquisition data of the same detection item under the target display screen ID, and generating a defect weight for the target display screen acquisition data according to a type analysis result of the defect feature label; performing an importance evaluation on the target display screen acquisition data according to the structural weight, the detection weight, and the defect weight; performing an importance evaluation on other display screen acquisition data in the new display screen acquisition data set, and integrating them to form a data importance evaluation result.

[0018] Optionally, the first analysis unit includes: obtaining a first freshness threshold, a second freshness threshold, a first importance score threshold and a second importance score threshold, the first freshness threshold is less than the second freshness threshold, and the first importance score threshold is less than the second importance score threshold; generating a freshness level for each display screen collected data in the display screen collected data set according to the data freshness evaluation result, the first freshness threshold and the second freshness threshold; generating an importance level for each display screen collected data in the display screen collected data set according to the data importance evaluation result, the first importance score threshold and the second importance score threshold; performing an update level analysis on each display screen collected data in the display screen collected data set according to the freshness level and the importance level, and generating an update level analysis result.

[0019] In a third aspect, an embodiment of the present application provides a device for dynamically updating a data set, including:

[0020] processor, memory, input and output units, and buses;

[0021] The processor is connected to the memory, input and output units, and the bus;

[0022] The memory stores a program, and the processor calls the program to execute the first aspect and any optional method of the first aspect.

[0023] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a program is stored. When the program is executed on a computer, the program executes the first aspect and any optional method of the first aspect.

[0024] It can be seen from the above technical solutions that this application has the following advantages:

[0025] This application first obtains a training data set for a detection model, where the detection model is an artificial intelligence model that will be added to the current display production line to perform display defect detection, and the training data set is the display data used by the detection model during pre-training; obtains a display acquisition data set in real time, where the display acquisition data set is data collected in the current display production line for display defect detection; performs a data freshness assessment on the display acquisition data set; performs a data importance assessment on the display acquisition data set; performs an update level analysis on the display acquisition data set based on the data freshness assessment results and the data importance assessment results; filters out a target data item set from the display acquisition data set based on the update level analysis results; and updates the filtered target data item set to the training data set.

[0026] By collecting real-time data from display screens on the production line, we obtain the latest data. We also conduct freshness and importance assessments on a large amount of display screen data. The evaluation results are used to filter and update the training data set, ensuring that the display screen defect detection model can continuously learn and adapt to new situations on the production line, thereby improving the data quality and model performance of model training. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the present application, the following briefly introduces the drawings required for use in the embodiments or descriptions of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0028] Figure 1 A schematic diagram of an embodiment of a method for dynamically updating a dataset of the present application;

[0029] Figure 2 A schematic diagram of an embodiment of a method for pre-training a detection model for this application;

[0030] Figure 3 A schematic diagram of an embodiment of a method for sorting display screen data collected by the present application;

[0031] Figure 4 A schematic diagram of an embodiment of a method for adjusting a display screen acquisition data set according to the present application;

[0032] Figure 5 A schematic diagram of an embodiment of a method for evaluating data freshness in this application;

[0033] Figure 6 A schematic diagram of an embodiment of a method for evaluating the importance of data in this application;

[0034] Figure 7 A schematic diagram of an embodiment of a method for updating grade analysis for this application;

[0035] Figure 8 This is a schematic diagram of an embodiment of the screen color correction device of the present application;

[0036] Figure 9 This is a schematic diagram of another embodiment of the screen color correction device of the present application. DETAILED DESCRIPTION

[0037] In the following description, specific details such as specific system structures and techniques are provided for purposes of illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obscuring the description of the present application with unnecessary detail.

[0038] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0039] It will also be understood that the term "and / or" used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0040] As used in this specification and the appended claims, the term "if" can be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0041] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0042] References to "one embodiment" or "some embodiments" in this specification mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in one or more embodiments of the present application. Thus, phrases such as "in one embodiment," "in some embodiments," "in other embodiments," and "in other embodiments" appearing in various places in this specification do not necessarily refer to the same embodiment, but rather mean "one or more but not all embodiments," unless otherwise specifically emphasized. The terms "including," "comprising," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0043] In the prior art, to address defect detection for new display screens, new artificial intelligence (AI) defect detection models are often added to the display screen production line. This approach allows for targeted defect detection while reducing production line change costs. After the detection model is designed and pre-deployment training is completed using a training dataset, it is often not yet ready for deployment. This is because the training data used by the detection model is often relatively old, and new display screens vary from batch to batch. Therefore, prior to deploying the detection model, fresher data is often required for training. Data freshness significantly impacts the performance of the detection model after deployment. However, existing training dataset update methods are often static, requiring manual selection of the latest batch of display screen data. Furthermore, because each display screen ID requires multiple images for defect detection within the same inspection project, the amount of display screen data collected is often very large, requiring additional screening of more important display screen data. This is time-consuming and inefficient, and prevents staff from accurately selecting display screen data from the current display screen production line that best matches the detection model, effectively reflecting the latest changes in the display screen data. This results in poor pre-deployment training results, reducing training data quality and detection model performance.

[0044] Based on this, the present application discloses a method, device and storage medium for dynamically updating a data set, which are used to improve the data quality and model performance of model training.

[0045] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0046] The method of the present application can be applied to a server, device, terminal or other device with logic processing capability, and the present application does not limit this. For the convenience of description, the following description is based on the example of the execution subject being a terminal.

[0047] See also Figure 1 The present application provides an embodiment of a method for dynamically updating a data set, including:

[0048] 101. Obtain a training data set for a detection model. The detection model is an artificial intelligence model that will be added to the current display production line to perform display defect detection. The training data set is display data used by the detection model during pre-training.

[0049] First, obtain a training dataset for the inspection model. This dataset should come from a variety of display production lines to ensure data diversity, including data from different models, production batches, and defect types. This ensures that the inspection model has sufficient display inspection capabilities during initial training.

[0050] 102. Acquire a display screen acquisition data set in real time. The display screen acquisition data set is data collected in a current display screen production line and used for display screen defect detection.

[0051] Next, we acquire display screen data sets in real time using high-precision, high-stability acquisition equipment already deployed on the production line. This ensures data quality to the greatest extent possible. Furthermore, we achieve real-time data collection and transmission, ensuring that the training model can learn the latest production line conditions in a timely manner.

[0052] Specifically, this step includes producing a client and a data source.

[0053] Production client: the production line where the model is deployed.

[0054] Data source: Equipment or sensors on the production line continuously collect new defect data, which directly reflects the latest status of the production line, including newly emerging defect types.

[0055] 103. Evaluate the data freshness of the display screen collected data set.

[0056] The core of the data freshness evaluation of the display screen collected data set is to determine the data collection time through the timestamp, and to evaluate the data freshness through data such as collection time, defect type, and frequency.

[0057] Specifically, this step includes a data freshness evaluation unit.

[0058] Data Freshness Assessment Unit: When new defect data is generated from the data source, it is sent to the Data Freshness Assessment Unit along with a timestamp. This unit is responsible for assessing the data's recency, or freshness. Freshness can be measured by the difference between the data's generation timestamp and the current time (i.e., Age of Information (AoI)), yielding a freshness rating.

[0059] 104. Conduct data importance assessment on the display screen collected data set.

[0060] Next, it is necessary to evaluate the importance of the data set collected by the display screen and conduct a multi-angle analysis of the data collected from each production batch.

[0061] Specifically, this step includes a data importance evaluation unit.

[0062] Data importance assessment unit: Evaluates the importance of each data item in the data source based on the data's contribution to the model performance, and uses the model's inference score on the data to measure and obtain the importance level.

[0063] 105. Perform update level analysis on the display screen collected data set based on the data freshness evaluation results and the data importance evaluation results.

[0064] Then, based on the data freshness and importance assessment results, the display screen data set is analyzed for update levels. A comprehensive score is calculated for each display screen data set, based on data freshness and importance. This score is then used to categorize the data into different levels, such as high priority, medium priority, and low priority.

[0065] 106. Filter out a target data item set from the display screen collected data set according to the update level analysis result.

[0066] Based on the update level analysis results, the target data item set is filtered from the display screen data set. High-priority data is prioritized, and when high-priority data is insufficient, medium-priority data is considered. This ensures that the amount of filtered data is moderate, neither too much resulting in inefficient processing nor too little resulting in inadequate model updates.

[0067] Specifically, this step includes a data screening unit.

[0068] Data screening unit: Filters out data items that need to be updated based on freshness and importance evaluation results.

[0069] 107. Update the filtered target data item set into the training data set.

[0070] Finally, the set of filtered target data items is updated to the training dataset, and the new data is integrated into the existing training dataset to ensure data coherence and consistency.

[0071] Specifically, this step includes a data updating unit.

[0072] Data update unit: adds the filtered data items to the data set to update the data set.

[0073] In an embodiment of the present application, a training data set for a detection model is obtained. The detection model is an artificial intelligence model that will be added to the current display production line to perform display defect detection. The training data set is the display data used during pre-training of the detection model. A display acquisition data set is acquired in real time. The display acquisition data set is data collected in the current display production line for display defect detection. A data freshness assessment is performed on the display acquisition data set. A data importance assessment is performed on the display acquisition data set. An update level analysis is performed on the display acquisition data set based on the data freshness and data importance assessment results. A target data item set is filtered from the display acquisition data set based on the update level analysis results. The filtered target data item set is updated into the training data set.

[0074] By automatically monitoring data changes, assessing data freshness and importance, and dynamically updating datasets, we improve the data quality and performance of model training. Specifically, by acquiring the latest data from displays on the production line in real time, we conduct freshness and importance assessments on a large amount of display data. The results of these assessments are then used to filter and update the training dataset. This ensures that the display defect detection model can continuously learn and adapt to new situations on the production line, improving the data quality and performance of model training.

[0075] Secondly, the embodiments of the present application also have the following beneficial effects:

[0076] 1. Automated operation: reduces manual intervention and improves the efficiency and accuracy of data updates.

[0077] 2. Improve data freshness: By dynamically updating the data set, the freshness of the data is improved, so that model training can reflect the latest changes in the data in a timely manner.

[0078] 3. Improve model performance: By training the model with the latest and most relevant data, the accuracy and generalization ability of the model are improved.

[0079] 4. Save resources: Optimize data screening strategies and reduce unnecessary data processing and storage requirements.

[0080] 5. Strong adaptability: This method can adapt to rapid changes in data and is suitable for various scenarios that require real-time updates of data sets.

[0081] See also Figure 2 , the present application provides an embodiment of a method for pre-training a detection model, comprising:

[0082] 201. Use the updated training dataset to pre-train the detection model before putting it into use.

[0083] 202. Evaluate the model performance of the detection model using the updated training dataset and generate performance feedback information.

[0084] 203. When the performance feedback information shows that the detection model performance meets the standards, the detection model is deployed to the display production line.

[0085] 204. Send performance feedback information of the detection model back to the production client for optimization in future model training.

[0086] In this embodiment of the present application, the terminal pre-trains the detection model using the updated training dataset before it is put into use. The purpose of pre-training is to enable the detection model to adapt to the updated dataset, learn new defect characteristics, and improve the model's generalization capabilities. Based on the complexity of the model and the size of the data, an appropriate training strategy is selected, such as gradually increasing the learning rate or using transfer learning.

[0087] Next, the terminal evaluates the detection model's performance using the updated training dataset. Specifically, appropriate evaluation metrics, such as precision, recall, F1 score, and AUC, are selected to comprehensively measure model performance. Cross-validation is also performed to assess the model's stability and generalization capabilities, preventing performance deviations caused by improper data partitioning. Performance benchmarks are compared with previous model versions to assess the extent of performance improvements or declines, as well as the new model's performance on specific defect types.

[0088] Once performance feedback indicates the inspection model meets performance requirements, the inspection model is deployed to the display production line. First, ensure that the model files, dependent libraries, and configuration files are properly packaged and deployed to the production environment. Compatibility testing is then performed within the production environment to ensure the model operates properly and seamlessly integrates with other system components. A monitoring and alarm system is deployed on the production line to track model performance changes in real time and promptly identify and address potential issues.

[0089] Finally, the terminal sends performance feedback from the detection model back to the production client for optimization in future model training. Specifically, the terminal collects model performance feedback from the production client, including accuracy, false positive rate, and missed negative rate. New defects appearing on the production line are annotated and added to the training dataset to enrich the model's training samples. Based on this performance feedback and the newly added training data, the model is iteratively trained and optimized to continuously improve its performance and accuracy. A closed-loop optimization mechanism is established to ensure that the model can continuously learn and adapt to new situations on the production line, forming a virtuous cycle of continuous improvement.

[0090] Through pre-training, performance evaluation, deployment, and feedback collection, a complete model update and optimization cycle is formed. During implementation, model performance changes are analyzed and training strategies and parameters are adjusted in a timely manner to ensure model stability and accuracy on the production line. Simultaneously, data labeling and collection processes are continuously optimized to enrich training samples and enhance the model's generalization capabilities. Furthermore, establishing an effective monitoring and alarm system is key to ensuring stable model operation. In future model training, new algorithms and technologies can be further explored to improve model performance and efficiency.

[0091] In an embodiment of the present application, the above steps are composed of a model training unit, a model evaluation unit, a model deployment unit and a performance feedback unit.

[0092] Model training unit: Use the updated dataset to perform model training;

[0093] Model evaluation unit: evaluates the performance of the model trained using the updated dataset;

[0094] Model deployment unit: If the model performance is good, the model is deployed to the production line;

[0095] Performance Feedback Unit: Sends model performance feedback information back to the production client for optimization in future model training. This feedback loop ensures that the system can continuously adapt to new defect types and changes.

[0096] See also Figure 3 The present application provides an embodiment of a method for sorting display screen collected data, comprising:

[0097] 301. Pre-train and sort the display screen collected data in the target data item set according to the update level analysis result.

[0098] In an embodiment of the present application, a training plan is provided for the detection model by updating the grade analysis results, and the display screen data that meets the conditions is sorted by freshness and importance, and the data with higher rankings are trained first to ensure that the detection model reaches the expectations before deployment more quickly.

[0099] See also Figure 4 The present application provides an embodiment of a method for adjusting a display screen acquisition data set, comprising:

[0100] 401. Screening out display screen collected data containing a circuit area from a display screen collected data set, and obtaining generation time data of the display screen collected data containing the circuit area.

[0101] In an embodiment of the present application, a plurality of display screen defect detection models are provided on a traditional display screen production line, mainly including a display screen flat area detection model, a display screen curved area detection model, etc., as well as an artificial intelligence model for detecting specific defects, a model for detecting screen body defects, and a model for detecting real defects.

[0102] The display screen acquisition data set is the image data collected on the above-mentioned production line. These image data are input into the display screen defect detection models for detection, and the corresponding defect detection data are obtained as sample annotations.

[0103] With the development of display screens, the functionality of display screens has gradually increased, and their structures have gradually become more complex, mainly reflected in the increase in the number of layers of the functional layers of the display screen, the degree of curvature of the screen, the splicing between the screens, and the addition of internal circuits between the functional layers. In the embodiments of the present application, the display screen with the added internal circuit is mainly discussed, and it is a thin circuit below the pixel layer (the thin circuit is located inside the display screen). The thin circuit is a circuit structure with a small thickness, much thinner than the PCB board below the display screen. It is specifically located below the pixel layer inside the display screen. While providing the corresponding pixel point display function for the display screen, the thinner thin circuit will also minimize the light reflection effect caused when the pixel layer is lit. However, as the function becomes more complex, the complexity of some thin circuits becomes higher, and the thickness inevitably approaches the critical value. For some display screens that are already very thin, it may reach the level of affecting the display of the pixel layer, and when the thin circuit realizes the pixel point function, some metal materials or alloys used have strong reflective ability. When a certain thickness is reached, it can reflect a specific light intensity. Although the human eye may not be able to easily detect the light reflection of the thin circuit inside the display screen, for detection instruments and defect detection models, this type of light reflection can affect the defect detection results.

[0104] Due to improvements in display screens, different thin circuits have been added beneath the pixel layer, creating distinct circuit regions on the display. This necessitates the generation of defect detection models specifically tailored to these regions within the existing production line. A certain number of displays with these regions have already been inspected on the production line, meaning a sufficient number of samples are available for pre-training. The goal is to analyze which of these display data sets are most suitable for pre-training the detection model to accelerate both accuracy and deployment. Using more important data allows for faster training, while using fresher data allows the detection model to better adapt to future displays (data from displays with more or more complex circuit regions). Traditional approaches typically simply screen out displays containing circuit regions and identify images with defects in these regions. This defect conclusion is generated by the existing defect detection model on the production line. The existing detection model's accuracy for these new defects is lower than that of the new defect detection model (the AI model designed to detect defects in the display circuit region) that will be added to the production line. However, this simple screening approach only identifies data from displays with very obvious defects in the circuit region. We believe that this type of data is of limited importance as training data for new defect detection models (AI models that detect defects in the display circuit area) before they are put into production lines. This can lead to poor pre-training effectiveness, prolonged training time, and low efficiency. To address this, we have designed a new method for screening data collected from display production lines to improve the quality and efficiency of pre-training for new defect detection models before they are put into production lines.

[0105] In this embodiment, the terminal selects display screen data containing circuit areas from the display screen data set. This is because display screen samples without circuit areas already exist in the original training data set and have undergone extensive training. These samples are also more numerous and relatively fresh. The terminal then obtains the generation time data for this type of display screen data containing circuit areas, which is used for subsequent freshness assessment.

[0106] 402. Perform defect area and defect type analysis on the display screen data collected containing the circuit area, and generate a defect feature label for each display screen data collected containing a defect in the circuit area.

[0107] In this embodiment of the present application, the terminal analyzes the defect area and defect type for data collected from display screens containing circuit regions, and generates a defect feature label for each display screen data sample containing defects in the circuit region. This is intended to prioritize data collected from display screens containing defects in the circuit region through detection using models in existing production lines. Because the existing defect detection models currently in operation on the production line may falsely or accurately detect defects in the circuit region, this type of data is of greater importance. The terminal will first label this type of display screen data with a defect feature label.

[0108] 403. Classify the display screen data collected after defect area and defect type analysis according to the display screen ID and the inspection item.

[0109] After analyzing defect areas and defect types, the terminal categorizes the collected display data by display ID and inspection project. This allows for the integration of multiple display data sets generated for the same inspection project for the same display ID into a single display data set. For the same display ID, multiple images are typically collected for one inspection project (one defect detection model). Even for the same grayscale image, at least two images are collected for repeated inspection.

[0110] 404. Generate a circuit structure hierarchy based on the classified display screen collected data, where the circuit structure hierarchy represents the structural level of the thin-film circuit in the circuit area.

[0111] Since each batch of display screens may not be the same, and because the display screen functions required by each manufacturer vary, the circuit area structure also varies. The thickness and reflectivity of the circuit area are the main factors affecting defect detection. The circuit structure hierarchy is composed of three factors: the area of the thin circuit, the thickness data (thickness gradient type, thickness uniform type), and the metal material. The larger the area, the higher the structure level; the more complex the thickness type, the higher the structure level (gradient type is more complex than uniform type); the larger the thickness, the higher the structure level; and the greater the metal reflectivity, the higher the structure level. Currently, the structural level is mainly divided manually, and the circuit structure hierarchy is divided accordingly. The circuit structure hierarchy is the same for the same model of display screen.

[0112] 405. Bind the acquired generation time data, defect feature labels, and circuit structure levels with the classified display screen collected data to generate a new display screen collected data set.

[0113] The terminal binds the acquired generation time data, defect feature labels, and circuit structure levels with the classified display screen acquisition data to generate a new display screen acquisition data set, that is, integrating the data of the same display screen ID and further organizing the data of the same inspection item. Finally, the generation time data, defect feature labels, and circuit structure levels are bound to the acquisition data of each display screen.

[0114] The beneficial effect of the embodiment of the present application is to organize data for subsequent freshness evaluation and importance evaluation, increase the accuracy of the evaluation, and screen out more important display screen collection data.

[0115] See also Figure 5 , the present application provides an embodiment of a method for evaluating data freshness, comprising:

[0116] 501. Select a target display screen collection data from a new display screen collection data set.

[0117] 502. Determine a time decay function based on defect detection items of data collected from a target display screen. The time decay function has an unknown number, namely, a generation time.

[0118] In an embodiment of the present application, the terminal selects a target display screen collection data from a new display screen collection data set. Next, the terminal determines a time decay function based on the defect detection items of the target display screen collection data. The time decay function has an unknown variable, the generation time. The defect detection items are mainly screen defect detection and display defect detection. Screen defect detection refers to the presence of structural defects in the screen, such as scratches, broken edges, etc. These defects are usually illuminated by a light source in an unlit state, and then the display screen is captured to capture an image. Display defect detection refers to defects that appear on pixel points, which need to be illuminated by inputting a calibration image, and then the image is captured and then detected. There are slight differences in the time decay functions of screen defect detection and display defect detection. The time decay function needs to be substituted into the generation time of the display screen collection data for calculation.

[0119] In the embodiment of the present application, the time decay function is first based on the characteristics of display screen production, because display screens are produced in batches, and batch defects will appear in the production process of display screens. Batch defects refer to the situation where a parameter or equipment (such as a robotic arm) on the production line is improperly set or accidentally improperly operated, resulting in a specific defect on the entire batch of display screens. The appearance of such defects is usually similar but there are differences in the location. For example, the robotic arm causes a deviation in the grasping position due to wear and tear, resulting in small screen scratch defects on the entire batch of display screens, but the screen scratch defects on each display screen may be different, in terms of location or degree of scratches. According to the characteristics of batch defects, a piecewise time decay function is usually selected, and the decay of the piecewise function is controlled to be faster, because such batch defects are usually small and not easy to observe. The cause is usually difficult to be quickly identified and eliminated, but once discovered, it is usually quickly eliminated. Therefore, the embodiment of the present application selects multiple means to evaluate the freshness of data, among which the time decay function is set.

[0120]

[0121] is the time decay function, is the attenuation coefficient, and This is a time parameter set based on the average troubleshooting time for batch defects in the current production line. is the data age, is the current time, The time when the data was generated.

[0122] 503. Generate a business weight based on the display screen model of the target display screen collected data. The business weight refers to the importance of the display screen corresponding to the display screen model to the training of the detection model.

[0123] Next, the terminal generates a service weight based on the display model of the target display. The service weight indicates the importance of the display model to the detection model training. Specifically, if the training dataset already contains data for some display models, the service weight for trained display models will be lower, while the service weight for untrained display models will be higher.

[0124] In this embodiment of the present application, the service weight is , only distinguishing the business weights of the trained display screens from the business weights of the untrained ones.

[0125] 504. Determine a data change rate parameter corresponding to the display screen model according to the display screen model of the target display screen data collection and the new display screen collection data set.

[0126] The terminal determines the data change rate parameter corresponding to the target display model based on the display model of the target display data collection data and the new display data collection data set. Specifically, the terminal determines the number of data in the display data collection data set that share the same "display model" as the target display data collection data, and generates a data change rate parameter based on this number and the total number of display data collection data sets. The data change rate refers to the size of the data flow of this type. If the flow is large, more training is required, while if the flow is small, less training is required.

[0127] 505. Freshness evaluation is performed based on the time decay function, the business weight, the data change rate parameter, and the generation time of the target display screen collected data.

[0128] In this embodiment, the freshness evaluation formula is as follows:

[0129]

[0130] in, For freshness evaluation formula, is the data change rate, 、 、 is the weighted value, the sum of the three is 1, The service weight of the current type of display screen.

[0131] 506. Perform freshness evaluation on other display screen collected data in the new display screen collected data set, and integrate them to form a data freshness evaluation result.

[0132] Finally, the terminal performs a freshness evaluation based on the time decay function, business weight and data change rate parameters, and the generation time of the target display screen collected data. Finally, the terminal performs a freshness evaluation on the other display screen collected data in the new display screen collected data set and integrates them to form the data freshness evaluation result.

[0133] This method utilizes a multi-dimensional comprehensive calculation of the freshness of display screen data. It uses a basic freshness calculation function (time decay function), business weight, and data flow to comprehensively calculate the freshness of the data, thereby providing greater freshness for the required data and improving the quality of display screen data collection and detection model performance.

[0134] See also Figure 6 , the present application provides an embodiment of a method for evaluating data importance, comprising:

[0135] 601. Generate a structure weight according to the circuit structure level of the target display screen collected data.

[0136] In an embodiment of the present application, the terminal generates a structural weight based on the circuit structure hierarchy of the data collected from the target display screen. The purpose is to incorporate complex circuit structures into the importance judgment. Because the more complex the circuit area structure, the higher its training value, it is necessary to generate a structural weight based on the reflectivity, thickness and other influencing factors of the circuit structure. The structural weight serves as an important item in the importance assessment.

[0137] 602. Determine the target display screen ID for collecting data of the target display screen, and analyze the defect feature label binding status of all display screen collected data of the same inspection item under the target display screen ID.

[0138] 603. Generate detection weights for the target display screen data collection based on the binding analysis results.

[0139] In the embodiment of the present application, the terminal determines the target display screen ID of the target display screen data, analyzes the defect feature label binding status of all display screen data collected for the same inspection project under the target display screen ID, and generates a detection weight for the target display screen data based on the binding analysis results. If some of the total display screen data collected for the same inspection project is bound to defect labels, but some is not, the importance of the portion without defect labels is adjusted.

[0140] For example, for display screen 001 (with a circuit area), 10 images were collected for screen defect detection. Defect analysis of these 10 images revealed defects in three images (images 01, 02, and 05). Subsequent analysis of the defect area and defect type revealed that image 01 detected a screen defect in the circuit area, while images 02 and 05 detected screen defects in the non-circuit area. A defect feature label was generated for image 01, and all 10 images were classified as part of the same batch. Image 01 was designated as a regular pre-training sample, while images 02 through 10 were designated as advanced training samples for the "new defect detection model." These samples are difficult for models from other production lines to detect, but are genuine examples of defects. It should be noted that if the above situation also occurs between the data corresponding to different display images within the same inspection project for the same display ID, the importance adjustment rules also apply.

[0141] Specifically, this part is to analyze the collected data of all the display screens with the same display screen ID and the same inspection project, and analyze whether there is data bound to defect feature labels in this group of data. If so, detection weights are generated for this group of data or only for the collected data of the target display screen.

[0142] 604. Analyze the defect feature labels of all display screen collected data of the same inspection item under the target display screen ID, and generate defect weights for the target display screen collected data according to the analysis results of the defect feature label types.

[0143] In an embodiment of the present application, the terminal analyzes the defect feature labels of all display screen collection data of the same detection project under the target display screen ID, and generates a defect weight for the target display screen collection data based on the type analysis results of the defect feature labels. Specifically, for example: Display screen No. 001 (with a circuit area) collected 10 images in the screen defect detection project, and defect analysis was performed on these 10 images. The detection results showed that there were defects in 3 images (No. 02 and No. 05). Both images have defects in the circuit area, and the defect types are different. No. 02 is defect A, and No. 05 is an unknown defect. The terminal then generates a defect weight for defect A for image No. 02. The defect weight is generated based on the detection accuracy of the defect in the current production line. If the defect is more difficult to detect, the defect weight is larger. An unknown defect means that the detection data of the current production line cannot be identified. The defect weight is generated manually.

[0144] 605. Evaluate the importance of collecting data for the target display screen based on the structure weight, detection weight, and defect weight.

[0145] 606. Perform importance evaluation on other display screen collected data in the new display screen collected data set, and integrate them to form a data importance evaluation result.

[0146] Finally, the terminal performs a significance assessment on the target display screen data based on the structural weight, detection weight, and defect weight. It then performs a significance assessment on the remaining display screen data in the new display screen data set, integrating the data significance assessment results. Specifically, this application adds the structural weight, detection weight, and defect weight to obtain the final significance assessment data.

[0147] This approach, by comprehensively considering the circuit area structure of the display data collected and analyzing the production line defect detection results, can more accurately assess the importance of each display image. This better assigns greater importance to the required data, improving the quality of the display data collected and the performance of the detection model.

[0148] See also Figure 7 , the present application provides an embodiment of a method for updating grade analysis, comprising:

[0149] 701. Obtain a first freshness threshold, a second freshness threshold, a first importance score threshold, and a second importance score threshold, wherein the first freshness threshold is smaller than the second freshness threshold, and the first importance score threshold is smaller than the second importance score threshold.

[0150] 702. Generate freshness levels for each piece of display acquisition data in the display acquisition data set according to the data freshness evaluation result, the first freshness threshold, and the second freshness threshold.

[0151] 703. Generate importance levels for each piece of display acquisition data in the display acquisition data set according to the data importance evaluation result, the first importance score threshold, and the second importance score threshold.

[0152] 704. Perform updated level analysis on each piece of display acquisition data in the display acquisition data set according to the freshness level and the importance level, and generate an updated level analysis result.

[0153] In the embodiment of the present application, the freshness evaluation unit uses the age of information (AoI) as the data freshness. In the present invention, the rules for dividing the data freshness levels are given:

[0154] ① Preset a first freshness threshold (AoI_1) and a second freshness threshold (AoI_2) according to the actual generation situation, where AoI_1 < AoI_2;

[0155] ② Compare the size relationship between the AoI of the current data item and AoI_1 and AoI_2 respectively;

[0156] ③ If AoI < AoI_1, the data freshness level is high; if AoI_1 < AoI < AoI_2, the data freshness level is medium; if AoI > AoI_2, the data freshness level is low;

[0157] The evaluation process of data freshness:

[0158] ① Data monitoring: The freshness evaluation unit first monitors the changes in the database, records the generation time, reception time, and other relevant parameters of the data;

[0159] ② Freshness calculation: Calculate the AoI freshness of each data item based on the timestamp of the data.

[0160] ③ Freshness level division: The evaluation unit compares the calculated AoI with the preset threshold to determine the data freshness level. A shorter AoI indicates that the data is fresher, while a longer AoI indicates that the data is not fresh or has become obsolete;

[0161] ④ Decision support: Use the evaluation result for data update decision-making, and preferentially select data items with a high freshness level for update.

[0162] The data importance evaluation unit is a key component in the method for dynamically updating a data set. Its purpose is to determine which data items in the data set have a significant impact on the performance of the model, so as to preferentially retain or update these data. The contribution degree of a data item to the model performance can be measured by its impact on the model evaluation metrics (such as accuracy, recall rate, F1 score, etc.).

[0163] In the embodiment of this application, the importance of data can be evaluated through a neural network model based on deep learning (a defect detection model that has been put into production on the production line). The specific grading rules are as follows:

[0164] 1. Preset the first importance score threshold Importance_Th1 and the second importance score threshold Importance_Th2 according to the actual generation situation, where 0 < Importance_Th1 < Importance_Th2 < 1;

[0165] 2. Conduct importance evaluation on the detection results of the data collected by the display screen through the neural network model of deep learning to obtain the corresponding classification score score. For example: if the detection model has learned this type of data, the classification score will be relatively high and the data importance will be relatively small. Otherwise, the classification score will be relatively low and the data importance will be relatively large. Therefore, the model inference score of the data is inversely proportional to the data importance score. It should be noted that the classification score score here is only one of the evaluation schemes and not the final importance evaluation result;

[0166] 3. Calculate the data importance score Importance_score, which can be calculated according to the model inference score score. For example, according to the case in the previous step, the data importance score can be calculated by the following formula: Importance_score = 1 - score. The final data importance score is proportional to the data importance;

[0167] 4. Compare the size relationship between the data importance score Importance_score and the first importance score threshold Importance_Th1 and the second importance score threshold Importance_Th2 respectively;

[0168] 5. If Importance_score < Importance_Th1, the data importance level is low; if Importance_Th1 < Importance_score < Importance_Th2, the data importance level is medium; if Importance_score > Importance_Th2, the data importance level is high;

[0169] Data importance assessment process:

[0170] ① Data preprocessing: Before evaluation, the data is preprocessed to ensure the accuracy of the evaluation;

[0171] ② Importance assessment: Evaluate the importance score of each data item based on a deep learning model;

[0172] ③ Importance level classification: Analyze the evaluation results and determine the importance level of the data item based on the pre-set importance threshold;

[0173] ④Decision support: Use the evaluation results for data update decisions, giving priority to updating data items with high importance levels.

[0174] Finally, the terminal performs an update level analysis on the data collected by each display screen in the display screen collection data set according to the freshness level and importance level, and generates an update level analysis result.

[0175] The data update conditions are determined by combining the data freshness level and the data importance level. The specific joint strategy is as follows:

[0176] High importance and high freshness: Data is both important and fresh and should be processed first;

[0177] High importance and medium freshness: The data is important but old, and the processing priority needs to be weighed;

[0178] High importance and low freshness: The data is important but too fresh, so it has a low processing priority;

[0179] Medium importance and high freshness: The data is generally important but fresh, and the processing priority needs to be weighed;

[0180] Medium importance and medium freshness: The data is generally important but relatively old, and the processing priority needs to be weighed;

[0181] Medium importance and low freshness: The data is generally important but too fresh, so the processing priority is low;

[0182] Low importance and high freshness: data is not important but fresh, so the processing priority is low;

[0183] Low importance and medium freshness: The data is not important but old, so the processing priority is low;

[0184] Low importance and low freshness: Data is not important and out of date, so it is not processed;

[0185] This partitioning approach helps achieve an optimal balance between value and freshness during data transmission and processing. This strategy allows for more efficient dataset updates and ensures timely feedback to the client about the latest dataset generation status. This ensures that the deployed model can adapt to emerging defect types on the production line, improving production line detection performance. It should be noted that the above method is only one way to update the level of analysis, and it offers the benefit of speed.

[0186] See also Figure 8 , the present application provides an embodiment of a screen color correction device, comprising:

[0187] The first acquisition unit 801 is used to obtain a training data set for a detection model. The detection model is an artificial intelligence model that will be added to the current display production line to perform display defect detection. The training data set is display data used during pre-training of the detection model.

[0188] The second acquisition unit 802 is used to acquire a display screen acquisition data set in real time. The display screen acquisition data set is data collected in the current display screen production line and used for display screen defect detection.

[0189] The third acquiring unit 803 is configured to filter out display screen acquisition data containing a circuit area from the display screen acquisition data set, and acquire generation time data of the display screen acquisition data containing the circuit area.

[0190] The second analysis unit 804 is configured to perform defect area and defect type analysis on the display screen data collected containing the circuit area, and generate a defect feature label for each display screen data collected containing a defect in the circuit area.

[0191] The classification unit 805 is used to classify the display screen data collected after defect area and defect type analysis according to the display screen ID and detection items.

[0192] The second generating unit 806 is configured to generate a circuit structure hierarchy according to the classified display screen collected data, where the circuit structure hierarchy represents a structural level of the thin-film circuit in the circuit area.

[0193] The third generating unit 807 is configured to bind the acquired generation time data, defect feature labels, and circuit structure levels with the classified display screen acquisition data to generate a new display screen acquisition data set.

[0194] The first evaluation unit 808 is configured to perform data freshness evaluation on the display screen collected data set.

[0195] Optionally, the steps of the first evaluation unit 808 include:

[0196] Select a target display screen from the new display screen acquisition data set to acquire data.

[0197] A time decay function is determined based on defect detection items of data collected from a target display screen. The time decay function has an unknown variable, namely, generation time.

[0198] A business weight is generated based on the display model of the target display screen's collected data. The business weight refers to the importance of the display screen corresponding to the display screen model to the training of the detection model.

[0199] The data change rate parameter corresponding to the display screen model is determined according to the display screen model of the target display screen data collection and the new display screen collection data set.

[0200] Freshness evaluation is performed based on the time decay function, business weight and data change rate parameters, and the generation time of the data collected by the target display screen.

[0201] Perform freshness evaluation on other display screen collected data in the new display screen collected data set and integrate them to form a data freshness evaluation result.

[0202] The second evaluation unit 809 is configured to perform data importance evaluation on the display screen collected data set.

[0203] Optionally, the steps of the second evaluation unit 809 include:

[0204] The structure weight is generated according to the circuit structure level of the target display screen acquisition data.

[0205] Determine the target display screen ID for target display screen data collection, and analyze the defect feature label binding status of all display screen data collected for the same inspection project under the target display screen ID.

[0206] Generate detection weights for target display screen acquisition data based on the binding analysis results.

[0207] Analyze the defect feature labels of all display screen collected data of the same inspection project under the target display screen ID, and generate defect weights for the target display screen collected data based on the type analysis results of the defect feature labels.

[0208] The importance of collecting data for the target display is evaluated based on the structure weight, detection weight, and defect weight.

[0209] The importance of other display screen collection data in the new display screen collection data set is evaluated and integrated to form a data importance evaluation result.

[0210] The first analyzing unit 810 is configured to perform an update level analysis on the display screen collected data set according to the data freshness evaluation result and the data importance evaluation result.

[0211] Optionally, the first analysis unit 810 includes:

[0212] A first freshness threshold, a second freshness threshold, a first importance score threshold, and a second importance score threshold are obtained, wherein the first freshness threshold is smaller than the second freshness threshold, and the first importance score threshold is smaller than the second importance score threshold.

[0213] A freshness level is generated for each display screen collected data in the display screen collected data set according to the data freshness evaluation result, the first freshness threshold and the second freshness threshold.

[0214] An importance level is generated for each piece of display screen collected data in the display screen collected data set according to the data importance evaluation result, the first importance score threshold, and the second importance score threshold.

[0215] An update level analysis is performed on each display screen collected data in the display screen collected data set according to the freshness level and the importance level, and an update level analysis result is generated.

[0216] The screening unit 811 is configured to screen out a target data item set from the display screen collected data set according to the update level analysis result.

[0217] The updating unit 812 is used to update the screened target data item set into the training data set.

[0218] The sorting unit 813 is configured to perform pre-training sorting on the display screen collected data in the target data item set according to the update level analysis result.

[0219] The pre-training unit 814 is used to use the updated training data set to pre-train the detection model before it is put into use.

[0220] The first generating unit 815 is used to evaluate the model performance of the detection model using the updated training data set and generate performance feedback information.

[0221] The deployment unit 816 is configured to deploy the detection model to the display production line when the performance feedback information shows that the detection model performance meets the requirements.

[0222] The feedback unit 817 is used to send performance feedback information of the detection model back to the production client for optimization in future model training.

[0223] See also Figure 9 , the present application provides a device for dynamically updating a data set, comprising:

[0224] Processor 901 , memory 902 , input / output unit 903 , and bus 904 .

[0225] The processor 901 is connected to the memory 902 , the input / output unit 903 , and the bus 904 .

[0226] The memory 902 stores a program, and the processor 901 calls the program to execute the following Figure 1 、 Figure 2 and Figure 3 、 Figure 4 、 Figure 5 、 Figure 6 and Figure 7 The method in .

[0227] The present application provides a computer-readable storage medium, wherein a program is stored on the computer-readable storage medium, and when the program is executed on a computer, the program performs the following operations: Figure 1 、 Figure 2 and Figure 3 、 Figure 4 、 Figure 5 、 Figure 6 and Figure 7 The method in .

[0228] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0229] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0230] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of the present application.

[0231] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0232] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, read-only memory), random access memory (RAM, random access memory), disk or optical disk, and other media that can store program code.

Claims

1. A method for dynamically updating a data set, characterized in that: include: Obtaining a training data set for a detection model, wherein the detection model is an artificial intelligence model that will be added to the current display production line to perform display defect detection. The training data set is display data used during pre-training of the detection model. The detection model is an artificial intelligence model that performs defect detection on the circuit area of the display screen. Acquire a display screen acquisition data set in real time, wherein the display screen acquisition data set is data collected in a current display screen production line for display screen defect detection; Screening out display screen collected data containing a circuit area from the display screen collected data set, and obtaining generation time data of the display screen collected data containing the circuit area; Analyze the defect area and defect type of the display screen data collected from the circuit area, and generate a defect feature label for each display screen data collected from the circuit area; After defect area and defect type analysis, the collected display data is classified according to the display ID and test items; generating a circuit structure hierarchy based on the classified display screen collected data, wherein the circuit structure hierarchy represents a structural level of a thin-sheet circuit in a circuit area; Bind the acquired generation time data, defect feature labels, and circuit structure levels with the classified display screen acquisition data to generate a new display screen acquisition data set; Performing data freshness evaluation on the display screen collected data set; The step of performing data freshness evaluation on the display screen collection data set includes: selecting a target display screen collection data from the new display screen collection data set; determining a time decay function based on the defect detection items of the target display screen collection data, wherein the time decay function has an unknown variable, namely, a generation time; generating a business weight based on the display screen model of the target display screen collection data, wherein the business weight refers to the importance of the display screen corresponding to the display screen model to the training of the detection model; determining a data change rate parameter corresponding to the display screen model based on the display screen model of the target display screen collection data and the new display screen collection data set; performing freshness evaluation based on the time decay function, the business weight, the data change rate parameter, and the generation time of the target display screen collection data; performing freshness evaluation on other display screen collection data in the new display screen collection data set, and integrating them to form a data freshness evaluation result; Performing data importance evaluation on the display screen collected data set; Performing an update level analysis on the display screen collected data set according to the data freshness evaluation results and the data importance evaluation results; Filtering a target data item set from the display screen collected data set according to the update level analysis result; Update the filtered target data item set to the training data set.

2. The method according to claim 1, characterized in that After the step of updating the screened target data item set into the training data set, the method further includes: Using the updated training data set to pre-train the detection model before putting it into use; Evaluating the model performance of the detection model using the updated training data set to generate performance feedback information; When the performance feedback information shows that the detection model performance meets the standards, the detection model is deployed to the display production line; The performance feedback information of the detection model is sent back to the production client for optimization in future model training.

3. The method according to claim 2, characterized in that After the step of updating the screened target data item set into the training data set and before the step of pre-training the detection model using the updated training data set before putting it into use, the method further includes: Pre-training and sorting are performed on the display screen collected data in the target data item set according to the update level analysis result.

4. The method according to claim 1, wherein The step of performing data importance evaluation on the display screen collected data set includes: Generate structure weights according to the circuit structure level of the target display screen acquisition data; Determine the target display screen ID for collecting data of the target display screen, and analyze the defect feature label binding status of all display screen collection data of the same inspection item under the target display screen ID; Generating a detection weight for the target display screen collected data according to the binding analysis result; Analyze the defect feature labels of all display screen collected data of the same inspection item under the target display screen ID, and generate defect weights for the target display screen collected data according to the type analysis results of the defect feature labels; Performing an importance assessment on the data collected for the target display screen according to the structure weight, the detection weight, and the defect weight; The importance of other display screen collection data in the new display screen collection data set is evaluated and integrated to form a data importance evaluation result.

5. The method according to any one of claims 1 to 4, characterized in that The step of performing update level analysis on the display screen collected data set according to the data freshness evaluation result and the data importance evaluation result includes: Obtaining a first freshness threshold, a second freshness threshold, a first importance score threshold, and a second importance score threshold, wherein the first freshness threshold is smaller than the second freshness threshold, and the first importance score threshold is smaller than the second importance score threshold; generating a freshness level for each display screen collected data in the display screen collected data set according to the data freshness evaluation result, the first freshness threshold, and the second freshness threshold; generating an importance level for each piece of display screen collected data in the display screen collected data set according to the data importance evaluation result, the first importance score threshold, and the second importance score threshold; An update level analysis is performed on each display screen collected data in the display screen collected data set according to the freshness level and the importance level, and an update level analysis result is generated.

6. A device for dynamically updating a data set, characterized in that: include: a first acquisition unit, configured to acquire a training data set for a detection model, wherein the detection model is an artificial intelligence model that will be incorporated into a current display production line to perform display defect detection, the training data set being display data used during pre-training of the detection model, and the detection model is an artificial intelligence model that performs defect detection on a display circuit area; A second acquisition unit is used to acquire a display screen acquisition data set in real time, wherein the display screen acquisition data set is data collected in the current display screen production line for display screen defect detection; A third acquiring unit is configured to filter out display screen acquisition data containing a circuit area from the display screen acquisition data set, and acquire generation time data of the display screen acquisition data containing the circuit area; The second analysis unit is used to analyze the defect area and defect type of the display screen data containing the circuit area, and generate a defect feature label for each display screen data containing a defect in the circuit area; A classification unit is used to classify the display screen data collected after defect area and defect type analysis according to the display screen ID and test items; A second generating unit is configured to generate a circuit structure hierarchy based on the classified display screen collected data, wherein the circuit structure hierarchy represents a structural level of the thin-film circuit in the circuit area; A third generating unit is used to bind the acquired generation time data, defect feature labels, circuit structure levels and the classified display screen acquisition data to generate a new display screen acquisition data set; A first evaluation unit, configured to perform data freshness evaluation on the display screen collected data set; A second evaluation unit is used to perform data importance evaluation on the display screen collected data set; A first analysis unit is configured to perform an update level analysis on the display screen collected data set according to a data freshness evaluation result and a data importance evaluation result; a screening unit, configured to screen out a target data item set from the display screen collected data set according to an update level analysis result; The updating unit is used to update the filtered target data item set into the training data set.

7. The device according to claim 6, characterized in that The device further comprises: A pre-training unit, configured to use the updated training data set to pre-train the detection model before it is put into use; A first generating unit is configured to evaluate the model performance of the detection model using the updated training data set and generate performance feedback information; a deployment unit, configured to deploy the detection model to a display production line when the performance feedback information indicates that the detection model performance meets the standards; The feedback unit is used to send performance feedback information of the detection model back to the production client for optimization in future model training.

8. A computer-readable storage medium having a program stored thereon, wherein the program, when executed on a computer, performs the method according to any one of claims 1 to 5.

Images