Model management method and device for re-judgment model, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a model management method and device for a re-judgment model, electronic equipment and a storage medium. The method comprises the steps of displaying a model management page, wherein the model management page displays at least one algorithm information item in one-to-one correspondence with at least one algorithm formula; in response to a model configuration instruction input by a user for any target algorithm information item in the at least one algorithm information item, displaying a model configuration window corresponding to the target algorithm information item; and in response to a model configuration operation of a user for each detection area of the product in the model configuration window, determining model configuration information corresponding to each detection area of the product, the model configuration information being used for indicating whether to start a re-judgment model for the corresponding detection area and / or a classification threshold of the corresponding detection area when the re-judgment model is started. According to the scheme, the user is allowed to perform targeted re-judgment model configuration for each detection area of the product, and the defects of the existing unified re-judgment scheme can be avoided.

Description

Technical Field

[0001] This invention relates to the field of product testing technology, and more specifically to a model management method and apparatus, electronic device and storage medium for a re-evaluation model. Background Technology

[0002] Product defect detection is a core aspect of quality control in manufacturing, aiming to identify various defects on or inside workpieces through automated means to ensure high yield rates. The following uses wafer defect detection as an example to describe existing problems in the technology. It should be understood that wafers are merely an example; other similar products requiring defect detection (such as LCD screen panels and lithium battery electrodes) also face the same issues. In semiconductor manufacturing processes, microscopic defects on the wafer surface directly affect chip yield and reliability. Wafer defect detection is a critical quality control step in semiconductor manufacturing, aiming to identify and locate minute defects on or inside wafers to ensure chip yield and reliability. The wafer detection process typically includes high-resolution imaging, image processing, and intelligent analysis. For example, high-contrast images of the wafer surface can be acquired, and defect features can be captured using line-scan or area-scan cameras under different lighting conditions. Subsequently, deep learning models or traditional image processing defect detection algorithms can be combined to classify and identify defects in the images.

[0003] In wafer inspection scenarios, various chip (die) designs and defect types are encountered. Traditional defect detection algorithms suffer from over-detection or under-detection due to factors such as the varying axis movement speed of imaging equipment. In other words, defects are diverse and variable, making accurate differentiation difficult using conventional algorithms. In such cases, a re-judgment model (e.g., an artificial intelligence model, or AI model) can be used to correct over-detection and identify new defects. Of course, besides wafers, other similar products may also require re-judgment based on their defect detection results.

[0004] In existing technical solutions, products may have different detection areas (or sub-regions, which can be represented by Zone), but the re-judgment model can only perform unified re-judgment on the entire product and cannot achieve targeted re-judgment processing for detection areas, thus failing to meet the detection needs of different detection areas. Summary of the Invention

[0005] The present invention is proposed in view of the above-mentioned problems. The present invention provides a model management method, a model management device, an electronic device, and a storage medium for a re-judgment model.

[0006] According to one aspect of the present invention, a model management method for a reassessment model is provided. The reassessment model is used to reassess the defect detection results of a product. The defect detection results are obtained through algorithm recipe detection of the product. The algorithm recipe includes at least one defect detection algorithm corresponding to at least one detection area of ​​the product. The method includes: displaying a model management page, which displays at least one algorithm information item corresponding to at least one algorithm recipe, the algorithm information item including first identification information of the corresponding algorithm recipe; responding to a model configuration instruction input by a user for any target algorithm information item among the at least one algorithm information item, displaying a model configuration window corresponding to the target algorithm information item, the algorithm recipe corresponding to the target algorithm information item being set with a corresponding reassessment model; responding to a user's model configuration operation for each detection area of ​​the product in the model configuration window, determining the model configuration information corresponding to each detection area of ​​the product, the model configuration information being used to indicate whether to enable the reassessment model for the corresponding detection area and / or, when enabling the reassessment model, a classification threshold for the corresponding detection area, wherein the classification threshold is a threshold used by the reassessment model for classifying when reassessing the defect detection results for the detection area.

[0007] For example, the model configuration window includes a first display area and a second display area. The first display area is used to display second identification information of each detection area of ​​the product, and the second display area is used to display a first input control for model configuration information. In response to a user's model configuration operation for each detection area of ​​the product in the model configuration window, determining the model configuration information corresponding to each detection area of ​​the product includes: in response to a user's first selection operation on the second identification information of any detection area of ​​the product, displaying a first input control for the model configuration information corresponding to that detection area in the second display area; and determining the model configuration information corresponding to that detection area based on the information entered by the user in the first input control. The model configuration operation includes a first selection operation and an input operation in the first input control.

[0008] For example, the model configuration window further includes a third display area for displaying a product image of the product; after displaying the model configuration window corresponding to the target algorithm information item, the method further includes: in response to a user's first selection operation on the second identification information of any detection area of ​​the product, displaying the product image in the third display area and highlighting the detection area on the product image.

[0009] For example, the model configuration window includes a global threshold application control. In response to the user's model configuration operation for each detection area of ​​the product in the model configuration window, the model configuration information corresponding to each detection area of ​​the product is determined, including: in response to the user's trigger operation on the global threshold application control, the preset global classification threshold or the global classification threshold input by the user is determined as the classification threshold corresponding to each of the product's detection areas; wherein, the model configuration operation includes the trigger operation on the global threshold application control.

[0010] For example, the model configuration window includes a fourth display area, which includes a second input control for a global classification threshold. In response to the user's model configuration operation for each detection area of ​​the product in the model configuration window, the window determines the model configuration information corresponding to each detection area of ​​the product. The window also includes determining a global classification threshold based on the information entered by the user in the second input control. The model configuration operation also includes an input operation in the second input control.

[0011] For example, in response to a user's triggering operation on the global threshold application control, the preset global classification threshold or the global classification threshold input by the user is determined as the classification threshold corresponding to each of the product's detection areas, including: in response to a user's triggering operation on the global threshold application control, outputting a prompt message, the prompt message being used to indicate the risk of applying the global classification threshold; and upon receiving a user's confirmation application instruction, determining the preset global classification threshold or the global classification threshold input by the user as the classification threshold corresponding to each of the product's detection areas.

[0012] For example, the algorithm information item further includes first setting information and second setting information of the corresponding algorithm recipe, the first setting information being used to indicate whether the corresponding re-judgment model has been set for the algorithm recipe, and the second setting information being used to indicate whether the corresponding model configuration information has been set for the algorithm recipe; and / or, the algorithm information item further includes a configuration start control, and the model configuration instruction includes a trigger operation on the configuration start control; and / or, the algorithm information item further includes third identification information of the machine that uses the corresponding algorithm recipe.

[0013] For example, the model management page includes a third input control, and the method further includes: searching for algorithm recipes that match the search information in a preset database based on the search information entered by the user in the third input control, so as to determine at least one algorithm recipe; wherein, the preset database is used to store the algorithm recipes of at least one product and the attribute information of each algorithm recipe, the attribute information including one or more of the following: first identification information of the algorithm recipe, third identification information of the machine using the algorithm recipe, and fourth identification information of the product corresponding to the algorithm recipe; the search information includes at least a portion of the information in the attribute information.

[0014] According to another aspect of the present invention, a model management device for a re-judgment model is also provided. The re-judgment model is used to re-judge the defect detection results of a product. The defect detection results are obtained through product algorithm recipe detection. The algorithm recipe includes at least one defect detection algorithm corresponding to at least one detection area of ​​the product. The device includes: a first display module for displaying a model management page, the model management page displaying at least one algorithm information item corresponding to at least one algorithm recipe, the algorithm information item including first identification information of the corresponding algorithm recipe; a second display module for displaying a model configuration window corresponding to the target algorithm information item in response to a model configuration command input by a user for any target algorithm information item in the at least one algorithm information item, the algorithm recipe corresponding to the target algorithm information item being set with a corresponding re-judgment model; and a determination module for determining the model configuration information corresponding to each detection area of ​​the product in response to a model configuration operation by a user in the model configuration window for each detection area of ​​the product, the model configuration information indicating whether to enable the re-judgment model for the corresponding detection area and / or the classification threshold of the corresponding detection area when the re-judgment model is enabled, wherein the classification threshold is the threshold used by the re-judgment model for classification when re-judging the defect detection results of the detection area.

[0015] According to another aspect of the present invention, an electronic device is also provided, including a processor and a memory, wherein the memory stores computer program instructions, which are executed by the processor to perform the model management method of the above-described re-judgment model.

[0016] According to another aspect of the present invention, a storage medium is also provided, on which program instructions are stored, wherein the program instructions are used to execute the model management method of the above-described re-judgment model when running.

[0017] The model management method, apparatus, electronic device, and storage medium of the re-judgment model according to embodiments of the present invention can provide users with a model management page. Users can manage the model configuration information of the re-judgment model corresponding to each algorithm formula in different detection areas of the product through the model management page. The model configuration information controls whether the re-judgment model is enabled in each detection area and / or the corresponding classification threshold when the re-judgment model is enabled. This model management scheme allows users to configure re-judgment models specifically for each detection area of ​​the product, facilitating users to enable or disable the re-judgment model in any detection area and / or set the classification threshold for any detection area as needed. This targeted re-judgment model management scheme avoids the drawbacks of existing unified re-judgment schemes, helps improve the efficiency of product defect detection, and / or avoids problems such as re-over-inspection of detection areas caused by the unified application of re-judgment models. Attached Figure Description

[0018] The above and other objects, features, and advantages of the present invention will become more apparent from the more detailed description of the embodiments of the invention in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same parts or steps.

[0019] Figure 1 A schematic flowchart illustrating a model management method for a review model according to an embodiment of the present invention is shown.

[0020] Figure 2 A schematic diagram illustrating a model management page according to an embodiment of the present invention is shown;

[0021] Figure 3 A schematic diagram of a model configuration window for a first detection region according to an embodiment of the present invention is shown;

[0022] Figure 4 A schematic diagram of a model configuration window for a second detection region according to an embodiment of the present invention is shown;

[0023] Figure 5 A schematic diagram of a model configuration window for a third detection region according to an embodiment of the present invention is shown;

[0024] Figure 6 A schematic block diagram of a model management device for a review model according to an embodiment of the present invention is shown; and

[0025] Figure 7 A schematic block diagram of an electronic device according to an embodiment of the present invention is shown. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the present invention more apparent, exemplary embodiments according to the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are merely a part of the embodiments of the present invention, and not all of the embodiments of the present invention. It should be understood that the present invention is not limited to the exemplary embodiments described herein. Based on the embodiments of the present invention described herein, all other embodiments obtained by those skilled in the art without inventive effort should fall within the protection scope of the present invention.

[0027] In existing technologies, once a review model is designed, it can be deployed in the product manufacturing site's inspection system or in the cloud. The local inspection system can then use the review model, either automatically or by downloading it from the cloud, to review the defect detection results of its products. The inspection system can acquire product images, perform defect detection using a preset defect detection algorithm, obtain defect detection results, and then use the review model to review those results. However, in existing technologies, the review model can only perform a uniform review of the entire product, failing to provide targeted review processing for specific inspection areas and thus not meeting the inspection needs of different areas. Research has found that this uniform review process leads to problems such as low inspection efficiency and re-inspection. For example, the defect detection results obtained by the preset defect detection algorithm are usually mostly accurate and do not require review; only a small number of results may have problems and require review. For instance, out of 100 inspection areas, only 10 areas may have defect detection results requiring review, while the remaining 90 areas do not. This overall review method is very inefficient, resulting in significant efficiency losses. Furthermore, because it is impossible to set different re-judgment methods (such as re-judgment classification thresholds) for each detection area, it is not possible to effectively and flexibly adapt the re-judgment to different defect types, which can easily lead to some detection areas passing the inspection again. In the existing technology, the product's re-judgment model cannot be managed and controlled by the user (such as the product manufacturer) who uses the re-judgment model. Therefore, the working method of the re-judgment model cannot be adjusted by the user, and a default uniform re-judgment method can only be adopted.

[0028] To at least partially address the aforementioned problems, embodiments of the present invention provide a model management method for re-judgment models. This model management scheme allows users to configure re-judgment models for each detection area of ​​a product, enabling or disabling re-judgment models for any detection area and / or setting classification thresholds for any detection area as needed. This targeted re-judgment model management scheme avoids the drawbacks of existing unified re-judgment schemes, helps improve the efficiency of product defect detection, and / or avoids problems such as re-over-inspection of detection areas caused by the unified application of re-judgment models.

[0029] Figure 1This document illustrates a schematic flowchart of a model management method 100 for a review model according to an embodiment of the present invention. The model management method 100 for the review model described herein can be applied to any electronic device with data processing and / or instruction execution capabilities, i.e., it is executed by an electronic device. This electronic device may include, but is not limited to, personal computers, servers, mobile terminals, etc. Exemplarily and not limitingly, the electronic device may include, for example, a testing system deployed at a product manufacturing site, i.e., the model management method 100 for the review model is executed by the testing system deployed at the product manufacturing site. The review model described herein is used to review the defect detection results of a product. The defect detection results are obtained through algorithmic formulation detection of the product. The algorithmic formulation includes at least one defect detection algorithm corresponding one-to-one with at least one detection area of ​​the product. The product described herein can be any type of product requiring defect detection, including but not limited to wafers, lithium battery electrodes, etc. The product may include one or more detection areas, and an algorithmic formulation may include one or more defect detection algorithms, with each defect detection algorithm corresponding one-to-one with a detection area. Any two detection areas of the same product may partially overlap or not overlap at all. Each detection area corresponds to a defect and is detected by the corresponding defect detection algorithm. The defect detection algorithm described in this paper can be any type of algorithm, including but not limited to deep learning models and / or traditional image processing algorithms. For example, defects may correspond to different detection regions. Figure 1 As shown, the model management method 100 includes steps S110, S120 and S130.

[0030] Step S110: Display the model management page. The model management page displays at least one algorithm information item that corresponds one-to-one with at least one algorithm recipe. The algorithm information item includes the first identification information of the corresponding algorithm recipe.

[0031] The electronic device used in the model management method 100 for performing the review model may include an output device or a communication connection to an output device. The communication connection described herein may include any wired and / or wireless connection. The output device may include any type of display device, which may be touch-sensitive or non-touch-sensitive. Any content that needs to be displayed can be displayed through the display device, such as a model management page. The model management page may be displayed automatically when preset conditions are met, or it may be displayed in response to a user's page opening operation. For example, preset conditions may include, but are not limited to, power-on, completion of product image acquisition, etc.

[0032] Figure 2This diagram illustrates a model management page according to an embodiment of the present invention. It should be noted that the positions of various UI elements such as display areas, pages, windows, and controls described herein can be arbitrarily set as needed and are not limited to the positions shown in the accompanying drawings. The model management page can display one or more algorithm information items corresponding one-to-one with one or more algorithm recipes. The algorithm information items are displayed in... Figure 2 The information shown is a single line of information under the recipe list. The algorithm information item can include attribute information related to the corresponding algorithm recipe, including but not limited to the algorithm recipe's first identifier. The first identifier can be any form of identifier, such as a number or name, as long as it can distinguish and identify the algorithm recipe. Figure 2 In the list, the first identifier is the recipe name; different algorithm recipes correspond to different recipe names. Users can select any row of information in the recipe list to choose the corresponding algorithm recipe. For example, Figure 2 The upper right corner of the model management page shown can display one or more of the following controls: maximize, minimize, and close (not shown in the figure). These controls are used to maximize, minimize, and close the model management page, respectively. Similarly, other windows and pages described in this article, such as the model configuration window, can also have one or more of these controls. The settings and functions of each control can be understood by referring to the model management page; they will not be elaborated upon here.

[0033] In step S120, in response to the model configuration instruction input by the user for any target algorithm information item in at least one algorithm information item, a model configuration window corresponding to the target algorithm information item is displayed, and the algorithm recipe corresponding to the target algorithm information item is set with a corresponding re-judgment model.

[0034] The electronic device used in the model management method 100 for executing the re-judgment model may include an input device or be communicatively connected to an input device. The input device may include, but is not limited to, one or more of a mouse, keyboard, touchpad, touchscreen, trackball, microphone, etc. Users can perform various operations through the input device, i.e., input various instructions and / or information. That is, the various user instructions described herein can be input in any way, including but not limited to input via triggering annotation controls on the user interface (UI), input via shortcut keys, input via voice, etc. It is understood that the aforementioned user interface is an interactive interface capable of displaying information; the model management page and various windows (such as the model configuration window) described herein can all be displayed on the user interface. Similarly, the various user operations described herein can also be implemented in any operation mode; when a user performs any operation, they input the corresponding instruction (or operation instruction). That is, the operations performed by the user may include, but are not limited to, triggering annotation controls on the user interface (UI), triggering shortcut keys, providing voice commands, etc.

[0035] The algorithm recipe corresponding to the target algorithm information item has a corresponding review model set, meaning that the review model can be configured for algorithm recipes with a review model set. Model configuration instructions are used to instruct the configuration of the algorithm recipe corresponding to the target algorithm information item. These instructions can be of any type, including but not limited to preset shortcut key input instructions and / or instructions entered by the user through triggering the configuration launch control on the model management page. The configuration launch control can be, for example,... Figure 2 The "Set AI De-detection Algorithm" button control shown is an example. The user clicks... Figure 2 The "Set AI De-detection Algorithm" button control shown can pop up a model configuration window corresponding to the currently selected algorithm recipe. It should be noted that the accompanying figures in this article use the AI ​​de-detection algorithm to represent the re-judgment model; however, the re-judgment model used for de-detection is only an example. When the re-judgment model is used to add defects, a model management method similar to that used in the embodiments of this invention can also be adopted.

[0036] In step S130, in response to the user's model configuration operation for each detection area of ​​the product in the model configuration window, the model configuration information corresponding to each detection area of ​​the product is determined. The model configuration information is used to indicate whether to enable the re-judgment model for the corresponding detection area and / or the classification threshold for the corresponding detection area when the re-judgment model is enabled. The classification threshold is the threshold used by the re-judgment model for classifying when re-judging the defect detection results for the detection area.

[0037] For each defect location indicated by the defect detection result, if the attribute value of the defect location belonging to any category label obtained by the re-judgment model for that defect location exceeds the classification threshold, then the label of the defect location is determined as that category label. The attribute value may include the score (representing probability) of the defect location belonging to that category label and / or the area of ​​the defect location. For example, the classification threshold may include a score threshold and / or an area threshold.

[0038] As described above, each product may include one or more detection areas. For each detection area, one or more defect detection algorithms can be used for detection. Each defect detection algorithm can detect one type of defect and obtain a corresponding defect location. Therefore, the defect detection result for each detection area may include one or more defect locations. Each defect location may specifically include one or more non-contiguous image regions. These image regions are determined by the defect detection algorithm to contain the defect it is trying to detect, such as particles, scratches, dents, etc. The re-judgment model has one or more category labels, which may include, but are not limited to, normal (ok), scratches, burrs, etc. The category labels used by the re-judgment model when classifying defects may be partially the same as, partially different from, completely the same as, or completely different from, the category labels used by the defect detection algorithm in the corresponding algorithm recipe when classifying defects. The category labels of the re-judgment model can be set as needed. For example, for defects that are classified as scratches by the defect detection algorithm, they can be ignored for some products, and the re-judgment model can classify such defects as normal. However, for some products, the impact is relatively large, and the re-judgment model can classify such defects as scratches. When reviewing defect detection results, the review model can classify each defect record according to its existing category labels. During classification, the review model can determine the score for each category label for the defect location in each defect record (each defect record corresponds to one defect location). The score represents the probability that the defect record belongs to a preset category label. If the score for any defect location belonging to any category label exceeds the score threshold of the review model, the review model can determine that the defect location belongs to the category indicated by that category label. Furthermore, by way of example and not limitation, the review model can also compare the image area occupied by the defect location with an area threshold. When the image area occupied by the defect location exceeds the area threshold, the review model determines that the defect location belongs to the category indicated by the preset defect category label used by the review model (in this case, the review model can be used to detect a single defect category, such as a scratch). Furthermore, by way of example and not limitation, the re-judgment model can compare the score corresponding to the defect location with a score threshold and compare the image area occupied by the defect location with an area threshold. When the score of any defect location belonging to any category label exceeds the score threshold of the re-judgment model and the image area occupied by the defect location exceeds the area threshold, the defect location is determined to belong to the category indicated by the category label.

[0039] Figure 3-5 The diagrams show the model configuration windows for different detection regions (which may be referred to as the first detection region, the second detection region, and the third detection region, respectively) according to embodiments of the present invention.

[0040] Figure 3The image shows the model configuration window for the first detection area (i.e., the surface detection area, Surface Zone). Figure 4 The image shown is the model configuration window for the second detection region (Zone2). Figure 5 The image shown is the model configuration window for the third detection region (Zone 10). It should be noted that the terms "first," "second," and "third," etc., used in this document are for distinguishing purposes only and have no other special meaning. Figure 3-5 As shown, the model configuration window displays the identification information of each detection area of ​​the product. Users can select the identification information of any detection area, that is, select the detection area, and set whether to enable the re-judgment model for the corresponding detection area and / or the classification threshold of the corresponding detection area when the re-judgment model is enabled.

[0041] By adopting the above technical solution, a model management page can be provided to users. Users can manage the model configuration information of the re-judgment model corresponding to each algorithm formula in different detection areas of the product through this page. The model configuration information controls whether the re-judgment model is enabled in each detection area and / or the corresponding classification threshold when the re-judgment model is enabled. This model management solution allows users to configure re-judgment models for each detection area of ​​the product, making it convenient for users to enable or disable the re-judgment model in any detection area and / or set the classification threshold for any detection area as needed. This targeted re-judgment model management solution avoids the drawbacks of existing unified re-judgment solutions, helps improve the efficiency of product defect detection, and / or avoids problems such as re-over-inspection in detection areas caused by the unified application of re-judgment models.

[0042] According to an embodiment of the present invention, the model configuration window includes a first display area and a second display area. The first display area is used to display second identification information of each detection area of ​​the product, and the second display area is used to display a first input control for model configuration information. In response to a user's model configuration operation for each detection area of ​​the product in the model configuration window, determining the model configuration information corresponding to each detection area of ​​the product includes: in response to a user's first selection operation on the second identification information of any detection area of ​​the product, displaying a first input control for the model configuration information corresponding to that detection area in the second display area; and determining the model configuration information corresponding to that detection area based on the information entered by the user in the first input control. The model configuration operation includes a first selection operation and an input operation in the first input control.

[0043] like Figure 3-5 As shown, the left side of the model configuration window is the first display area, and the lower right side is the second display area. Figure 3 In the diagram, the first display area 310 and the second display area 320 are marked with dashed boxes. Figure 4-5The first and second display areas are not clearly marked, but according to Figure 3 The positions of the first and second display areas are understandable. Of course, this is understandable. Figure 3-5 This is merely an example; the positions of the first and second display areas shown, as well as the specific information they contain, can be set as needed and are not limited to this example. Figure 3-5 The location and specific information of each detection area of ​​the product. The first display area is used to display the second identification information of each detection area. Similar to the first identification information, the second identification information can be any form of identification information such as number or name, as long as it can distinguish and identify the detection area. The first selection operation can be any type of selection operation, as long as it can select any detection area to be configured in the model from the detection areas. For example, the first selection operation can include, but is not limited to, clicking to select the second identification information of the detection area in the detection area list or detection area drop-down menu control by means of a mouse or touch screen, or selecting the second identification information of the detection area in the detection area list or detection area drop-down menu control by a preset shortcut key, or entering the second identification information of the detection area in the input control (which can be called the fourth input control), etc. Figure 3-5 As shown, the first display area on the left displays the region names of each detection zone. When the user selects the second identification information of any detection zone, the second display area on the right displays the first input control of the model configuration information corresponding to the currently selected detection zone (shown as "current zone" in the figure). The first input controls corresponding to any two different detection zones can be partially the same and partially different, completely the same, or completely different, depending on the needs of defect re-judgment for the detection zone. It can be understood that when the detection zone selected by the user changes, the model configuration information displayed in the second display area on the right can change synchronously. For example, the first input control may include an enable control for controlling the activation of the re-judgment model and / or a threshold setting control for setting the classification threshold of the re-judgment model. Figure 3-5 In the example, the enable control is shown as a checkbox, labeled "Enable AI?" to the left. It's important to understand that the checkbox is just an example; the enable control could be other types of controls, such as a toggle switch.

[0044] In addition, Figure 3-5 The text box also shows an input text box for the classification threshold. For example, the classification threshold may include a score threshold and / or an area threshold. In the second display area, the user can set the aforementioned score threshold and / or area threshold as needed.

[0045] In the model configuration information, default settings can be used for information not set by the user. For example, if the user has not enabled the control controls, enabling the control controls can default to disabling the review model. If the user has not set the score threshold and / or area threshold, the score threshold and / or area threshold can use default values. These default values ​​can be preset default values, or global classification thresholds set by the user using the control through the following global thresholds.

[0046] Using the above technical solution, the model configuration window can display the second identification information of each detection region, making it convenient for users to select the detection region to be configured. Furthermore, after the user selects a detection region, the model configuration window can also display the first input control, allowing the user to set the corresponding model configuration information for that detection region. This visible interactive method facilitates quick selection of detection regions and setting of model configuration information, improving the efficiency and interactivity of model management and effectively enhancing the user experience.

[0047] According to an embodiment of the present invention, the model configuration window further includes a third display area for displaying a product image of the product; after displaying the model configuration window corresponding to the target algorithm information item, the method further includes: in response to a user's first selection operation on the second identification information of any detection area of ​​the product, displaying the product image in the third display area and highlighting the detection area on the product image.

[0048] See Figure 3-5 The product image is displayed in the middle area of ​​the model configuration window. The currently selected detection area can be highlighted on each product image. The highlighting method can be any method that allows the detection area to be identified from the product image, including but not limited to highlighting the entire detection area's mask with a preset color distinct from other image areas, or highlighting the outline of the detection area with a preset color distinct from other image areas. For example, in... Figure 3 In the product image shown, the currently selected first detection area (Surface Zone) 330 is outlined with a rectangular bounding box. Figure 4 In the image shown, the currently selected second detection region (Zone2) 410 is outlined with a circular bounding box. Figure 5 In the image shown, the currently selected third detection region (Zone10) 510 is outlined with a rectangular bounding box.

[0049] By adopting the above solution, after the user selects any detection area, the detection area can be highlighted on the product image in real time, so that the user can quickly confirm and understand the location of the currently selected detection area, which helps to improve the efficiency and accuracy of model management and further enhances the user experience.

[0050] According to an embodiment of the present invention, the model configuration window includes a global threshold application control. In response to a user's model configuration operation for each detection area of ​​the product in the model configuration window, the model configuration information corresponding to each detection area of ​​the product is determined, including: in response to a user's trigger operation on the global threshold application control, a preset global classification threshold or a global classification threshold input by the user is determined as the classification threshold corresponding to each of the product's detection areas; wherein, the model configuration operation includes a trigger operation on the global threshold application control.

[0051] like Figure 3-5 As shown, the upper right corner of the model configuration window displays the global classification threshold setting area (i.e., the fourth display area). Figure 3 In the middle, the fourth display area 340 is marked with a dashed box. Figure 4-5 The fourth display area is not explicitly marked, but according to Figure 3 The location of the fourth display area is understandable. In the model configuration window, the fourth display area is marked "Global Threshold". Within this area, a "Synchronize Threshold to All Sub-regions" button control is displayed; this button control is the global threshold application control. When the user clicks this button control (i.e., executes the trigger operation for the global threshold application control), the preset global classification threshold or the user-input global classification threshold can be set as the corresponding classification threshold for each of the product's detection areas. Figure 3-5 The document also shows a setting control for the global classification threshold (i.e., the second input control described below), allowing the user to set the global classification threshold. However, it should be noted that this is only an example, and the global classification threshold can also be a preset fixed threshold. After applying the global classification threshold to each detection area, the user can optionally continue to adjust the global classification threshold based on the first input control corresponding to any detection area to obtain a personalized classification threshold for that detection area.

[0052] By employing the above technical solution, a global threshold application control allows users to apply global classification thresholds to each detection region. This facilitates the rapid, batch setting of classification thresholds for each detection region. Since products typically have a large number of detection regions, the global threshold application method significantly reduces the workload for users in setting classification thresholds, effectively improving model management efficiency.

[0053] According to an embodiment of the present invention, the model configuration window includes a fourth display area, the fourth display area including a second input control for a global classification threshold, responding to the user's model configuration operation for each detection area of ​​the product in the model configuration window, determining the model configuration information corresponding to each detection area of ​​the product, and further including: determining a global classification threshold based on the information input by the user in the second input control; wherein, the model configuration operation also includes an input operation in the second input control.

[0054] like Figure 3-5 As shown, the fourth display area may also include a second input control for the global classification threshold. The global classification threshold may include a score threshold and / or an area threshold. Users can easily set and adjust the global classification threshold through the second input control. Compared with a preset global classification threshold, this approach provides users with greater freedom, allowing them to adjust the size of the global classification threshold as needed.

[0055] According to an embodiment of the present invention, in response to a user's trigger operation on the global threshold application control, determining a preset global classification threshold or a user-input global classification threshold as the classification threshold corresponding to each of the product's detection areas includes: in response to a user's trigger operation on the global threshold application control, outputting a prompt message, the prompt message being used to indicate the risk of applying the global classification threshold; and upon receiving a user-input confirmation instruction, determining the preset global classification threshold or the user-input global classification threshold as the classification threshold corresponding to each of the product's detection areas.

[0056] Applying a global classification threshold to each detection area will override the original classification thresholds of each detection area. Therefore, to avoid the risk of a significant increase in subsequent workload due to incorrect overriding, a prompt message can be output when the user triggers the global threshold application control to inform the user of this risk, rather than applying it directly. For example, the prompt message can include, but is not limited to, one or more of text, image, and voice information. For instance, the prompt message can be output through a separate pop-up window. If the user confirms the application (i.e., enters the confirmation command), the operation of setting the preset global classification threshold or the user-input global classification threshold as the corresponding classification threshold for each of the product's detection areas can be executed. The confirmation command can be entered via a button control or shortcut key. For example, "OK" and "Cancel" button controls can be displayed in the pop-up window containing the prompt message. If the user clicks the "OK" control, the operation of setting the preset global classification threshold or the user-input global classification threshold as the corresponding classification threshold for each of the product's detection areas can be executed. If the user clicks the "Cancel" control, the operation of setting the preset global classification threshold or the user-input global classification threshold as the corresponding classification threshold for each of the product's detection areas will not be executed.

[0057] By employing the above technical solution, a prompt message can be output before applying the global classification threshold, thus avoiding the risk of increased user workload due to erroneous overriding caused by applying the global classification threshold. This solution helps reduce user errors and improves the accuracy of model management.

[0058] According to an embodiment of the present invention, the algorithm information item further includes first setting information and second setting information of the corresponding algorithm recipe, wherein the first setting information is used to indicate whether the corresponding re-judgment model has been set for the algorithm recipe, and the second setting information is used to indicate whether the corresponding model configuration information has been set for the algorithm recipe; and / or, the algorithm information item further includes a configuration start control, wherein the model configuration instruction includes a trigger operation on the configuration start control; and / or, the algorithm information item further includes third identification information of the machine using the corresponding algorithm recipe.

[0059] For example, the algorithm information item also includes first setting information and second setting information for the corresponding algorithm recipe. The first setting information indicates whether the algorithm recipe has set the corresponding re-judgment model, and the second setting information indicates whether the algorithm recipe has set the corresponding model configuration information. (Return to Reference) Figure 2 On the model management page, each algorithm information item can also include first and second settings for the algorithm recipe. The first setting is the information in the "Whether the model has been trained" column. If the algorithm recipe has a corresponding re-judgment model set, it will display "Trained Model"; otherwise, it will display "Untrained Model". The second setting is "Whether AI de-detection has been set". If the algorithm recipe has a corresponding model configuration set, it will display "Not Set"; otherwise, it will display "Not Set". Through these first and second settings, the setting status of the re-judgment model for each algorithm recipe can be clearly displayed to the user, facilitating the management of re-judgment models for different algorithm recipes.

[0060] For example, the algorithm information item also includes a configuration launch control, and the model configuration instructions include a triggering operation on the configuration launch control. See also Figure 2 Each algorithm information item can also include a configuration startup control, namely a "Set AI to bypass detection algorithm" button control. As mentioned above, the user clicks... Figure 2 The "Set AI Over-detection Algorithm" button control shown can pop up a model configuration window corresponding to the currently selected algorithm formula. This visual configuration launch control allows users to easily and quickly start configuring the review model, reducing the burden on users to remember shortcut keys or other operation methods.

[0061] For example, the algorithm information item also includes third identification information of the machine using the corresponding algorithm formula. See [link to relevant documentation] Figure 2Each algorithm information item can also include a third identification information of the machine that uses the corresponding algorithm formula, namely the "machine number". Machine errors can lead to a large number of products processed by that machine having similar defects. Therefore, by displaying the third identification information, users can quickly identify the machine on which each algorithm formula is applied, thereby quickly finding the machine that needs defect re-judgment and configuring the re-judgment model for that machine's algorithm formula.

[0062] For example, in the embodiment where "the algorithm information item also includes first setting information and second setting information of the corresponding algorithm recipe, the first setting information is used to indicate whether the algorithm recipe has set the corresponding re-judgment model, and the second setting information is used to indicate whether the algorithm recipe has set the corresponding model configuration information," and in the embodiment where "the algorithm information item also includes a configuration start control, and the model configuration instruction includes a trigger operation on the configuration start control," and in the embodiment where "the algorithm information item also includes third identification information of the machine using the corresponding algorithm recipe," any two or three of these can be combined together. The implementation methods and technical effects of the combined embodiments can be understood by referring to the descriptions of the various embodiments above, and will not be repeated here.

[0063] According to an embodiment of the present invention, the model management page includes a third input control, and the method further includes: searching for algorithm formulas that match the search information in a preset database based on search information input by the user in the third input control, so as to determine at least one algorithm formula; wherein, the preset database is used to store algorithm formulas for at least one product and attribute information of each algorithm formula, the attribute information including one or more of the following: first identification information of the algorithm formula, third identification information of the machine using the algorithm formula, and fourth identification information of the product corresponding to the algorithm formula; the search information includes at least a portion of the information in the attribute information.

[0064] Each algorithm recipe can have corresponding attribute information, which may include one or more of the following: a first identifier of the algorithm recipe, a third identifier of the machine using the algorithm recipe, and a fourth identifier of the product corresponding to the algorithm recipe; the search information includes at least a portion of the attribute information. Users can enter search information in the third input control to search for matching algorithm recipes from a preset database and display them on the model management page. For example... Figure 2 As shown, the upper left corner of the model management page displays a "Recipe Name" text box and a "Search" button control. Users can enter the recipe name (i.e., the first identifier information) of the algorithm recipe to be managed in the "Recipe Name" text box and click the "Search" button control to search for matching algorithm recipes from the preset database. The algorithm information items of the searched algorithm recipes are then displayed in a list on the model management page. Note that... Figure 2The “Recipe Name” text box shown is just an example; you can also search based on other information, such as the third identification information of the machine mentioned above.

[0065] By adopting the above technical solution, users can quickly search for the required algorithm formulas to manage the re-judgment model as needed. This solution can further improve the efficiency of model management and user experience.

[0066] According to another aspect of the present invention, a model management device for a re-judgment model is provided. The re-judgment model is used to re-judge the defect detection results of a product. The defect detection results are obtained by detecting the product through an algorithm formula. The algorithm formula includes at least one defect detection algorithm that corresponds one-to-one with at least one detection area of ​​the product. Figure 6 A schematic block diagram of a model management device 600 for reviewing models according to an embodiment of the present invention is shown. Figure 6 As shown, the device 600 may include a first display module 610, a second display module 620, and a determination module 630.

[0067] The first display module 610 is used to display a model management page. The model management page displays at least one algorithm information item that corresponds one-to-one with at least one algorithm recipe. The algorithm information item includes the first identification information of the corresponding algorithm recipe.

[0068] The second display module 620 is used to respond to the model configuration command input by the user for any target algorithm information item in at least one algorithm information item, and to display the model configuration window corresponding to the target algorithm information item, wherein the algorithm recipe corresponding to the target algorithm information item is set with a corresponding re-judgment model.

[0069] The determination module 630 is used to respond to the user's model configuration operation for each detection area of ​​the product in the model configuration window, and to determine the model configuration information corresponding to each detection area of ​​the product. The model configuration information is used to indicate whether to enable the re-judgment model for the corresponding detection area and / or the classification threshold for the corresponding detection area when the re-judgment model is enabled. The classification threshold is the threshold used by the re-judgment model for classifying when re-judging the defect detection results for the detection area.

[0070] According to another aspect of the present invention, an electronic device is also provided. Figure 7 A schematic block diagram of an electronic device 700 according to an embodiment of the present invention is shown, such as Figure 7 As shown, the electronic device 700 may include a processor 710 and a memory 720. The memory 720 stores a computer program, and the processor 710 executes the computer program to implement the aforementioned model management method for the re-evaluation model.

[0071] According to another aspect of the present invention, a storage medium is also provided. This medium stores a computer program / instructions, which, when executed by a processor, implement the model management method for the aforementioned review model. The storage medium may, for example, include a storage component of a tablet computer, a hard disk of a personal computer, a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a portable compact disc read-only memory (CD-ROM), a USB memory, or any combination of the above storage media. The computer-readable storage medium may be any combination of one or more computer-readable storage media.

[0072] Those skilled in the art can understand the specific implementation schemes and beneficial effects of the above-mentioned model management device, electronic device and storage medium for the review model by reading the relevant description of the model management method for the review model. For the sake of brevity, they will not be described in detail here.

[0073] Although exemplary embodiments have been described herein with reference to the accompanying drawings, it should be understood that the above exemplary embodiments are merely illustrative and are not intended to limit the scope of the invention. Various changes and modifications can be made therein by those skilled in the art without departing from the scope and spirit of the invention. All such changes and modifications are intended to be included within the scope of the invention as claimed in the appended claims.

[0074] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0075] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed.

[0076] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0077] Similarly, it should be understood that, in order to streamline the invention and aid in understanding one or more of the various aspects of the invention, features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof in the description of exemplary embodiments of the invention. However, this approach should not be construed as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim. Rather, as reflected in the corresponding claims, its inventive point lies in solving the corresponding technical problem with fewer features than all of those in a single disclosed embodiment. Therefore, the claims following the detailed description are hereby expressly incorporated into that detailed description, wherein each claim itself is a separate embodiment of the invention.

[0078] Those skilled in the art will understand that, apart from the mutual exclusion of features, all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or units of any method or apparatus so disclosed can be combined in any combination. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0079] Furthermore, those skilled in the art will understand that although some embodiments described herein include certain features but not others included in other embodiments, combinations of features from different embodiments are intended to be within the scope of the invention and form different embodiments. For example, in the claims, any of the claimed embodiments can be used in any combination.

[0080] The various component embodiments of the present invention can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will understand that microprocessors or digital signal processors (DSPs) can be used in practice to implement some or all of the functions of some modules in the model management device for the re-judgment model according to embodiments of the present invention. The present invention can also be implemented as an apparatus program (e.g., a computer program and computer program product) for performing part or all of the methods described herein. Such programs implementing the present invention can be stored on a computer-readable medium or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

[0081] It should be noted that the above embodiments are illustrative of the invention and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names.

[0082] The above description is merely a specific embodiment of the present invention or an explanation of that embodiment. The scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A model management method for a re-judgment model, characterized in that, The re-judgment model is used to re-judge the defect detection results of the product. The defect detection results are obtained through algorithmic formula detection of the product. The algorithmic formula includes at least one defect detection algorithm corresponding one-to-one with at least one detection area of ​​the product. The method includes: The model management page is displayed, which displays at least one algorithm information item that corresponds one-to-one with at least one algorithm recipe. The algorithm information item includes the first identification information of the corresponding algorithm recipe. In response to a model configuration command input by a user for any target algorithm information item among the at least one algorithm information items, a model configuration window corresponding to the target algorithm information item is displayed, wherein the algorithm recipe corresponding to the target algorithm information item is set with a corresponding re-judgment model; In response to the user's model configuration operation for each detection area of ​​the product in the model configuration window, the model configuration information corresponding to each detection area of ​​the product is determined. The model configuration information is used to indicate whether to enable the re-judgment model for the corresponding detection area and / or the classification threshold for the corresponding detection area when the re-judgment model is enabled. The classification threshold is the threshold used by the re-judgment model for classification when re-judging the defect detection results for the detection area.

2. The method according to claim 1, characterized in that, The model configuration window includes a first display area and a second display area. The first display area is used to display the second identification information of each detection area of ​​the product, and the second display area is used to display the first input control of the model configuration information. The step of responding to the user's model configuration operation for each detection area of ​​the product in the model configuration window, and determining the model configuration information corresponding to each detection area of ​​the product, includes: In response to a user's first selection operation of the second identification information of any detection area of ​​the product, a first input control for the model configuration information corresponding to the detection area is displayed in the second display area; Based on the information entered by the user in the first input control, determine the model configuration information corresponding to the detection area; The model configuration operation includes the first selection operation and the input operation in the first input control.

3. The method according to claim 2, characterized in that, The model configuration window also includes a third display area, which is used to display a product image of the product. After displaying the model configuration window corresponding to the target algorithm information item, the method further includes: In response to a user's first selection operation of second identification information for any detection area of ​​the product, an image of the product is displayed in the third display area and the detection area is highlighted on the product image.

4. The method according to any one of claims 1-3, characterized in that, The model configuration window includes a global threshold application control. In response to a user's model configuration operation for each detection area of ​​the product in the model configuration window, the system determines the model configuration information corresponding to each detection area of ​​the product, including: In response to the user's trigger operation on the global threshold application control, the preset global classification threshold or the global classification threshold input by the user is determined as the classification threshold corresponding to each of the detection areas of the product. The model configuration operation includes triggering the application of the global threshold control.

5. The method according to claim 4, characterized in that, The model configuration window includes a fourth display area, which includes a second input control for the global classification threshold. The step of determining the model configuration information corresponding to each detection area of ​​the product in response to a user's model configuration operation for each detection area of ​​the product in the model configuration window further includes: The global classification threshold is determined based on the information entered by the user in the second input control; The model configuration operation also includes input operations in the second input control.

6. The method according to claim 4, characterized in that, The step of responding to a user's trigger operation on the global threshold application control, determining a preset global classification threshold or a user-input global classification threshold as the classification threshold corresponding to each of the product's detection areas, includes: In response to the user's triggering operation on the global threshold application control, a prompt message is output, which is used to indicate the risk of applying the global classification threshold; Upon receiving a user-input confirmation instruction, the preset global classification threshold or the user-input global classification threshold is determined as the classification threshold corresponding to each of the product's detection areas.

7. The method according to any one of claims 1-3, characterized in that, The algorithm information item further includes first setting information and second setting information corresponding to the algorithm recipe. The first setting information is used to indicate whether the algorithm recipe has been set with the corresponding re-judgment model, and the second setting information is used to indicate whether the algorithm recipe has been set with the corresponding model configuration information; and / or The algorithm information item also includes a configuration startup control, and the model configuration instructions include a trigger operation on the configuration startup control; and / or, The algorithm information item also includes third identification information of the machine that uses the corresponding algorithm formula.

8. The method according to any one of claims 1-3, characterized in that, The model management page includes a third input control, and the method further includes: Based on the search information entered by the user in the third input control, an algorithm recipe that matches the search information is searched in a preset database to determine the at least one algorithm recipe; The preset database is used to store the algorithm formula of at least one product and the attribute information of each algorithm formula. The attribute information includes one or more of the following: the first identification information of the algorithm formula, the third identification information of the machine using the algorithm formula, and the fourth identification information of the product corresponding to the algorithm formula; the search information includes at least some of the information in the attribute information.

9. A model management device for a re-judgment model, characterized in that, The re-judgment model is used to re-judge the defect detection results of the product. The defect detection results are obtained through algorithmic formula detection of the product. The algorithmic formula includes at least one defect detection algorithm corresponding one-to-one with at least one detection area of ​​the product. The device includes: The first display module is used to display a model management page, wherein the model management page displays at least one algorithm information item corresponding to at least one algorithm recipe, and the algorithm information item includes the first identification information of the corresponding algorithm recipe; The second display module is used to respond to a model configuration instruction input by the user for any target algorithm information item in the at least one algorithm information item, and to display a model configuration window corresponding to the target algorithm information item, wherein the algorithm recipe corresponding to the target algorithm information item is set with a corresponding re-judgment model; The determination module is used to respond to the user's model configuration operation for each detection area of ​​the product in the model configuration window, and to determine the model configuration information corresponding to each detection area of ​​the product. The model configuration information is used to indicate whether to enable the re-judgment model for the corresponding detection area and / or the classification threshold for the corresponding detection area when the re-judgment model is enabled. The classification threshold is the threshold used by the re-judgment model for classification when re-judging the defect detection results for the detection area.

10. An electronic device comprising a processor and a memory, wherein, The memory stores computer program instructions, which, when executed by the processor, are used to perform the model management method of the review model as described in any one of claims 1-8.

11. A storage medium on which program instructions are stored, wherein, The program instructions are used to execute the model management method of the re-judgment model as described in any one of claims 1-8 when the program is running.

Images