Defect detection method, device, equipment and computer readable medium

Abstract

The present application provides a defect detection method, apparatus, device and computer-readable medium. By acquiring an image to be detected of an object to be detected, the image to be detected is input into a pre-trained defect detection model to obtain the defect detection category output by the defect detection model and the category confidence corresponding to each defect detection category; according to each defect detection category and the category confidence corresponding to each defect detection category, a candidate defect category is determined; a wafer feature map is acquired, and a wafer defect category is determined based on the wafer feature map; and a target defect category of the image to be detected is determined according to the candidate defect category and the wafer defect category. It can at least be used to solve the technical problem of inaccurate semiconductor defect positioning.

Description

Technical Field

[0001] The present application relates to the field of semiconductor technology, and in particular to a defect detection method, device, equipment and computer-readable medium. Background Art

[0002] The semiconductor production process has many steps and is extremely sophisticated and complex. Any problem in any small link may greatly affect production capacity. How to more accurately analyze defect data and determine the correct category of defects has become a key link in quickly locating the real cause of the defects. However, the inventors have found that there are at least the following technical problems in the relevant technology: the current defect category analysis method only analyzes the photos taken of each defect and then classifies them. Although this can solve some problems, many defect categories have high similarity in pictures, but the causes are very different. For such defects, it is difficult to determine the real cause by only analyzing the defect pictures. Engineers need to participate in investigating the causes many times, which consumes human resources. Summary of the invention

[0003] One object of the present application is to provide a defect detection method, apparatus, device and computer-readable medium, at least to solve the technical problem of inaccurate semiconductor defect positioning.

[0004] To achieve the above objectives, some embodiments of the present application provide the following aspects:

[0005] In a first aspect, some embodiments of the present application further provide a defect detection method, including:

[0006] Acquire an image to be detected of the object to be detected, input the image to be detected into a pre-trained defect detection model, and obtain defect detection categories output by the defect detection model and category confidences corresponding to each defect detection category;

[0007] Determining candidate defect categories according to each of the defect detection categories and the category confidence corresponding to each of the defect detection categories;

[0008] Obtaining a wafer feature map, and determining a wafer defect category based on the wafer feature map;

[0009] A target defect category of the image to be inspected is determined according to the candidate defect categories and the wafer defect category.

[0010] In a second aspect, some embodiments of the present application further provide a defect detection device, including:

[0011] The detection model prediction module is used to obtain an image to be detected of the object to be detected, input the image to be detected into a pre-trained defect detection model, and obtain the defect detection category output by the defect detection model and the category confidence corresponding to each defect detection category;

[0012] A candidate category determination module, used to determine a candidate defect category according to each defect detection category and a category confidence corresponding to each defect detection category;

[0013] A wafer defect classification module, used to obtain a wafer feature map and determine a wafer defect category based on the wafer feature map;

[0014] The target category determination module is used to determine the target defect category of the image to be detected according to the candidate defect category and the wafer defect category.

[0015] In a third aspect, some embodiments of the present application further provide a computer device, characterized in that the device comprises: one or more processors; and a memory storing computer program instructions, wherein the computer program instructions, when executed, cause the processor to execute the method described above.

[0016] In a fourth aspect, some embodiments of the present application further provide a computer-readable medium having computer program instructions stored thereon, wherein the computer program instructions can be executed by a processor to implement the method as described above.

[0017] Compared with the prior art, the solution provided by the embodiment of the present application obtains the image to be detected of the object to be detected, inputs the image to be detected into the pre-trained defect detection model, obtains the defect detection category output by the defect detection model and the category confidence corresponding to each defect detection category; determines the candidate defect category according to each defect detection category and the category confidence corresponding to each defect detection category; obtains the wafer feature map, determines the wafer defect category based on the wafer feature map; determines the target defect category of the image to be detected according to the candidate defect category and the wafer defect category. Combining the two dimensions for defect location improves the accuracy of defect location. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic diagram of a defect detection method provided in an embodiment of the present application;

[0019] Figure 2 A schematic diagram of a wafer feature map provided in an embodiment of the present application;

[0020] Figure 3 A schematic diagram of a wafer problem area provided in an embodiment of the present application;

[0021] Figure 4 A schematic diagram of another wafer problem area provided in an embodiment of the present application;

[0022] Figure 5 A schematic diagram of a flow chart of another defect detection method provided in an embodiment of the present application;

[0023] Figure 6 A schematic diagram of the structure of a defect detection device provided in an embodiment of the present application;

[0024] Figure 7 A schematic diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0026] Embodiment 1

[0027] Figure 1 A flowchart of a defect location method provided in an embodiment of the present application. This embodiment can be used for situations where defects are located in semiconductors. The method can be performed by a defect location device, which can be implemented in the form of hardware and / or software, such as the defect location device can be configured in a computer device. Figure 1 As shown, the method includes:

[0028] Step S101, obtaining an image to be detected of the object to be detected, inputting the image to be detected into a pre-trained defect detection model, and obtaining a defect detection category output by the defect detection model and a category confidence corresponding to each defect detection category.

[0029] In this embodiment, defect detection is performed by combining defect information and wafer information to improve the accuracy of defect detection.

[0030] In this step, the defect information itself is identified by a pre-trained defect detection model. Optionally, the defect detection model is used to perform model recognition on the image to be detected of the object to be detected to obtain prediction information output by the defect detection model, wherein the prediction information output by the defect detection model may include defect detection categories and category confidence of each defect detection category.

[0031] It is understandable that the defect detection category in the prediction information output by the defect detection model is determined based on the annotation information in the model training sample. Considering the complexity of the semiconductor production process, many defect information have high similarity, but the causes of the defects are very different. This situation often requires manual intervention to determine the cause of the defect. In order to avoid the above situation, in this application, when labeling the sample, not only the defect category corresponding to the sample is labeled, but on the basis of the defect category of the labeled sample, the similar category of the defect category is also labeled as an associated category. Exemplarily, taking five categories A, B, C, D, and E as examples, assuming that category B is similar to category A, and a sample is category B, then the sample can be labeled as B_A. If category C is similar to D and E, the sample of category C can be labeled as C_D_E. The above labeling method enables the trained defect detection model to output the same category as the labeled category, making the prediction of the defect detection model more accurate.

[0032] Optionally, before the image to be detected is input into a pre-trained defect detection model, the method further includes: obtaining a model training sample, training the defect detection model based on the model training sample, and obtaining a trained defect detection model. The defect detection model can be constructed with reference to an existing neural grid model, which is not limited here. The training of the defect detection model can also refer to an existing model training method, which is not described in detail here. The model training sample can be constructed by annotating the sample using the above-mentioned annotation method.

[0033] Step S102: determining candidate defect categories according to each of the defect detection categories and the category confidence corresponding to each of the defect detection categories.

[0034] In order to make the defect category predicted based on the model more accurate, in an embodiment of the present application, multiple candidate defect categories are determined through the prediction results output by the model.

[0035] In some embodiments of the present application, candidate defect categories may be directly screened out from each defect detection category based on the category confidence corresponding to each defect detection category, or may be screened out from each defect detection category based on the association between the defect detection categories.

[0036] In some embodiments of the present application, determining the candidate defect category according to each defect detection category and the category confidence corresponding to each defect detection category includes:

[0037] sorting the defect detection categories in reverse order based on the category confidence;

[0038] Traversing is performed according to the reverse sorting order, and for the traversed current defect detection category, the candidate defect category is determined according to the category attribute information of the current defect detection category and the category confidence of the current defect detection category, until the traversal end condition is met.

[0039] It is understandable that the higher the category confidence, the greater the probability of the corresponding defect detection category. Based on this, the defect detection categories predicted and output by the model can be sorted in reverse order based on the category confidence. The higher the category confidence of the defect detection category, the higher the ranking. Then, the reverse sorting order is used as the traversal order, and each defect detection category is analyzed in turn to determine the candidate defect category. When the traversal end condition is reached, the traversal is ended, and the candidate defect category when the traversal end condition is reached is used as the final candidate defect category.

[0040] Exemplarily, assuming that the predicted output of the model includes N defect detection categories C1, C2, C3, C4, C5...Ci, Cj, and the confidences of the defect detection categories are arranged in reverse order such that P(Ci)>P(Cj)..., then the traversal starts from the defect detection category Ci corresponding to P(Ci), and for the defect detection category Ci, the candidate defect category is determined according to the category attribute information of Ci and the category confidence of Ci, and it is judged whether the traversal end condition is met. When the traversal end condition is met, the current candidate defect category is directly used as the final candidate defect category. When the traversal end condition is not met, for the next defect detection category Cj of Ci, the operation of Ci is repeated, and a new candidate defect category is added to the candidate defect category. This is repeated until the traversal end condition is met, and the candidate defect category at the end of the traversal is used as the final candidate defect category.

[0041] For the current defect detection category traversed, the candidate defect category is determined according to the category attribute information of the current defect detection category and the category confidence of the current defect detection category, which can be: determining the candidate category determination rule according to the category attribute information of the current defect detection category, and determining the candidate defect category according to the candidate category determination rule. Among them, the category attribute information of the defect detection category can be an associated category, or an unassociated category, and its specific attribute information can be determined according to the annotation information (such as the category identifier of the defect detection category, etc.). When the category attribute information of the current defect detection category is an associated category, the current defect detection category and its associated categories can be directly used as candidate detection categories; when the category attribute information of the current defect detection category is an unassociated information, the current defect detection category is used as a candidate detection category, and at the same time, it is determined whether the next defect detection category meets the set conditions, and when the next defect detection category meets the set conditions, the next defect detection category is used as a candidate detection category.

[0042] Still taking the above prediction output as an example, assuming that the current defect detection category is Ci, if Ci does not have a similar category (this information can be obtained through category annotation), if and only if P(Cj) satisfies It is considered that Cj can be used as a candidate defect category.

[0043] In some embodiments of the present application, the traversal end condition includes at least one of the following:

[0044] The relationship between the category confidence of the current defect detection category and the category confidence of the next defect detection category does not satisfy the set condition;

[0045] The number of candidate defect categories meets the set threshold.

[0046] The functional relationship between the category confidence of the current defect detection category and the category confidence of the next adjacent defect detection category can be set as a set condition. When the relationship between the category confidence of the current defect detection category and the category confidence of the next defect detection category does not meet the set condition, it is determined that the similarity between the next defect detection category and the current defect detection category is low, and the traversal is stopped; and / or, a threshold value of the number of candidate defect categories is set as a set threshold value. When the number of candidate defect categories meets the set threshold value, the traversal is stopped. By setting the traversal end condition, the candidate defect categories are made more accurate.

[0047] Step S103, obtaining a wafer feature map, and determining a wafer defect category based on the wafer feature map.

[0048] In this step, wafer information is identified through the wafer feature map to identify the wafer defect category. Through deep learning, feature analysis and other methods, the possible defects of the wafer itself can be obtained on a macro scale, and the location of the defects on the wafer can be locked to obtain the wafer problem area and wafer defect category. There may be multiple problem areas on the wafer. If this happens, the category judgment and location determination are performed separately.

[0049] In some embodiments of the present application, determining the wafer defect category based on the wafer feature map includes:

[0050] According to the defect point distribution in the wafer feature map, the wafer problem area is determined, and the wafer defect category is determined based on the pre-constructed standard point distribution of the standard defect category and the defect point distribution.

[0051] The standard point distribution corresponding to the standard defect category can be determined based on the distribution characteristics of the known wafer defect category, and the association relationship between the standard defect category and the standard point distribution can be established. When identifying the wafer defect category in the wafer feature map, the standard defect category corresponding to the defect point distribution identified in the wafer feature map is determined as the wafer defect category according to the pre-established association relationship.

[0052] It is understandable that the standard defect categories can be updated as the defect categories are identified, and the corresponding standard point distributions can also be updated and enriched accordingly.

[0053] The defect point information is displayed in the wafer feature map, and the wafer problem area can be directly determined based on the defect point distribution, such as directly taking the area with dense defect point distribution as the wafer problem area. Figure 2 A schematic diagram of a wafer feature map provided in an embodiment of the present application, Figure 2 The points in the wafer feature map shown represent points where defects may exist. The areas where the points are concentrated are areas where there is a greater possibility of problems in the semiconductor production process, and can be regarded as wafer problem areas.

[0054] There may be one or more wafer problem areas in the wafer feature map. Figure 3 A schematic diagram of a wafer problem area provided in an embodiment of the present application; Figure 4 A schematic diagram of another wafer problem area provided in an embodiment of the present application. Figure 3 The area with dense dots in the wafer feature map shown is only the strip area at the edge of the wafer feature map, that is, Figure 3 The wafer feature map shown shows only a single wafer problem area Di. Figure 4 There are two independent densely populated areas in the wafer feature map shown: the circular area in the lower left corner of the wafer feature map and the strip area at the edge. Figure 4 There are two wafer problem areas in the wafer characterization map shown.

[0055] Step S104 , determining a target defect category of the image to be inspected according to the candidate defect categories and the wafer defect category.

[0056] After the candidate defect categories and the wafer defect categories are determined, the target defect category is determined by combining the candidate defect categories and the wafer defect categories.

[0057] In some embodiments of the present application, when the number of the wafer problem area is one, determining the target defect category of the image to be inspected according to the candidate defect category and the wafer defect category includes:

[0058] If there is a category in the candidate defect categories that is consistent with the wafer defect category, the wafer defect category is used as the target defect category;

[0059] If there is no category consistent with the wafer defect category among the candidate defect categories, the category with the highest category confidence among the candidate defect categories is used as the target defect category.

[0060] If there is only a single problem area on the wafer, when there is a category in the candidate defect categories that is consistent with the wafer defect category, there is no need to consider whether the wafer defect category corresponds to the wafer problem area, and the wafer defect category is directly used as the target defect category. When there is no category in the candidate defect categories that is consistent with the wafer defect category, a defect category is selected from the candidate defect categories as the target defect category. The higher the category confidence, the greater the probability. Based on this, the candidate defect category with the largest category confidence is selected as the target defect category.

[0061] In some embodiments of the present application, when the number of the wafer problem areas is at least two, determining the target defect category of the image to be inspected according to the candidate defect category and the wafer defect category includes:

[0062] The target defect category is determined according to the wafer problem area, the wafer defect category and the candidate defect categories.

[0063] If there are multiple problem areas on the wafer, there may be multiple situations, and different rules need to be used to determine the target defect category based on different situations.

[0064] Optionally, determining the target defect category according to the wafer problem area, the wafer defect category, and the candidate defect category includes:

[0065] If the wafer defect category corresponds to the wafer problem area, and there is a category consistent with the wafer defect category among the candidate defect categories, then the wafer defect category is used as the target defect category;

[0066] If the wafer defect category does not correspond to the wafer problem area and there is no defect category corresponding to the wafer problem area in the candidate defect categories, or the wafer defect category corresponds to the wafer problem area and there is a defect category corresponding to the wafer problem area in the candidate defect categories, then the category confidence of the candidate defect category is updated based on the position information of the candidate defect category, and the candidate defect category with the highest updated category confidence is used as the target defect category;

[0067] If the wafer defect category does not correspond to the wafer problem area, and there is no defect category corresponding to the wafer problem area in the candidate defect categories, the candidate defect category with the highest category confidence is used as the target defect category.

[0068] When there are multiple problem areas on a wafer, there are four possible scenarios:

[0069] ① The wafer defect category corresponds to the wafer problem area, and there is a category consistent with the wafer defect category in the candidate defect categories, and the category is directly output as the final result.

[0070] ② When the wafer defect category does not correspond to the wafer problem area and there is no defect category corresponding to the wafer problem area in the candidate defect categories, the diameter of the wafer is regarded as unit 1, and the Euclidean distance d between the location of the micro defect and the closest problem area of ​​the same category is calculated (after removing the noise corresponding to the problem area), and the confidence for the category is updated P(Cj) = P(Cj)*1 / d, and so on. After recalculating the confidence, arrange it in descending order, and the category corresponding to the highest confidence after sorting is the final result.

[0071] ③ When the wafer defect category corresponds to the wafer problem area and there is a defect category corresponding to the wafer problem area among the candidate defect categories, the logic in ② is used to recalculate the result and the highest value is taken as the final result.

[0072] ④ When the wafer defect category does not correspond to the wafer problem area, and there is no defect category corresponding to the wafer problem area in the candidate defect categories, the category corresponding to the highest value in the current confidence is directly output as the final result.

[0073] The solution provided by the embodiment of the present application obtains the image to be detected of the object to be detected, inputs the image to be detected into a pre-trained defect detection model, obtains the defect detection category output by the defect detection model and the category confidence corresponding to each defect detection category; determines the candidate defect category according to each defect detection category and the category confidence corresponding to each defect detection category; obtains the wafer feature map, determines the wafer defect category based on the wafer feature map; determines the target defect category of the image to be detected according to the candidate defect category and the wafer defect category. It can at least be used to solve the technical problem of inaccurate semiconductor defect positioning.

[0074] Embodiment 2

[0075] Figure 2It is a flow chart of another defect location method provided by an embodiment of the present application. This embodiment provides a preferred embodiment based on the above embodiment. This embodiment takes into account that the defects are caused by the parameter settings in the process or the influence of the production equipment, and these influences will be reflected on the wafer. Using the idea of ​​Image Pyramid, the information on the wafer (macro) + the information of the defect itself (micro) are jointly classified, and a weighted average method is used to achieve a better classification effect, which can improve the robustness of the classification algorithm to a certain extent.

[0076] Figure 2 The complete process of sample labeling, model training, model prediction, wafer feature map analysis, and final target defect category determination is schematically shown in FIG.

[0077] like Figure 2 As shown in the figure, after the samples are labeled with similar defect categories, model training is performed, and then the trained model is used to predict the defect category to obtain the defect detection category output by the model and the confidence of each defect detection category, and the confidence is arranged in descending order: P(Ci)>P(Cj)..., and the candidate defect categories are screened out as the multi-label output results; at the same time, the wafer feature map is analyzed to obtain the wafer defect category and position, and based on deep learning and feature analysis methods, the possible defects of the wafer itself can be obtained on a macro scale, and the position of the defect on the wafer can be locked; finally, the multi-label output results and the wafer defect category and position are combined to determine the final target defect category output.

[0078] Among them, the method for determining the candidate defect category, the method for determining the wafer defect category, and the method for determining the target defect category can refer to the above embodiments and will not be repeated here.

[0079] The embodiment of the present application performs a joint analysis and joint classification of defects generated in semiconductor production from different scales combined with different information, which can further improve the accuracy of defect classification, especially for some defects with high similarity that are difficult to distinguish. It provides a more accurate evaluation. This directly reduces the frequency of engineers' review of algorithm classification results, further improves work efficiency, and further enhances the accuracy of the analysis of the causes of defects.

[0080] Embodiment 3

[0081] Figure 3 Schematic diagram of a defect location device provided in an embodiment of the present application. Figure 3 As shown, the device comprises:

[0082] The detection model prediction module 310 is used to obtain an image to be detected of the object to be detected, input the image to be detected into a pre-trained defect detection model, and obtain the defect detection category output by the defect detection model and the category confidence corresponding to each defect detection category;

[0083] A candidate category determination module 320, configured to determine a candidate defect category according to each defect detection category and a category confidence corresponding to each defect detection category;

[0084] A wafer defect classification module 330 is used to obtain a wafer feature map and determine a wafer defect category based on the wafer feature map;

[0085] The target category determination module 340 is used to determine the target defect category of the image to be detected according to the candidate defect categories and the wafer defect category.

[0086] The solution provided by the embodiment of the present application is to obtain the image to be detected of the object to be detected through the detection model prediction module 310, input the image to be detected into the pre-trained defect detection model, obtain the defect detection category output by the defect detection model and the category confidence corresponding to each defect detection category; the candidate category determination module 320 determines the candidate defect category according to each defect detection category and the category confidence corresponding to each defect detection category; the wafer defect category module 330 obtains the wafer feature map and determines the wafer defect category based on the wafer feature map; the target category determination module 340 determines the target defect category of the image to be detected according to the candidate defect category and the wafer defect category. Combining the two dimensions for defect location improves the accuracy of defect location.

[0087] Optionally, based on the above solution, the wafer defect classification module 330 is specifically used for:

[0088] According to the defect point distribution in the wafer feature map, the wafer problem area is determined, and the wafer defect category is determined based on the pre-constructed standard point distribution of the standard defect category and the defect point distribution.

[0089] Optionally, based on the above solution, the candidate category determination module 320 is specifically used to:

[0090] sorting the defect detection categories in reverse order based on the category confidence;

[0091] Traversing is performed according to the reverse sorting order, and for the traversed current defect detection category, the candidate defect category is determined according to the category attribute information of the current defect detection category and the category confidence of the current defect detection category, until the traversal end condition is met.

[0092] Optionally, based on the above solution, the traversal end condition includes at least one of the following:

[0093] The relationship between the category confidence of the current defect detection category and the category confidence of the next defect detection category does not satisfy the set condition;

[0094] The number of candidate defect categories meets the set threshold.

[0095] Optionally, based on the above solution, when the number of the wafer problem area is one, the target category determination module 340 is specifically used to:

[0096] If there is a category in the candidate defect categories that is consistent with the wafer defect category, the wafer defect category is used as the target defect category;

[0097] If there is no category consistent with the wafer defect category among the candidate defect categories, the category with the highest category confidence among the candidate defect categories is used as the target defect category.

[0098] Optionally, based on the above solution, when the number of the wafer problem areas is at least two, the target category determination module 340 is specifically used to:

[0099] The target defect category is determined according to the wafer problem area, the wafer defect category and the candidate defect categories.

[0100] Optionally, based on the above solution, the target category determination module 340 is specifically used to:

[0101] If the wafer defect category corresponds to the wafer problem area, and there is a category consistent with the wafer defect category among the candidate defect categories, then the wafer defect category is used as the target defect category;

[0102] If the wafer defect category does not correspond to the wafer problem area and there is no defect category corresponding to the wafer problem area in the candidate defect categories, or the wafer defect category corresponds to the wafer problem area and there is a defect category corresponding to the wafer problem area in the candidate defect categories, then the category confidence of the candidate defect category is updated based on the position information of the candidate defect category, and the candidate defect category with the highest updated category confidence is used as the target defect category;

[0103] If the wafer defect category does not correspond to the wafer problem area, and there is no defect category corresponding to the wafer problem area in the candidate defect categories, the candidate defect category with the highest category confidence is used as the target defect category.

[0104] The defect detection device provided in the embodiment of the present invention can execute the defect detection method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0105] Embodiment 4

[0106] In addition, the embodiment of the present application also provides a computer device, Figure 4 A schematic diagram of the structure of a computer device provided in an embodiment of the present application, the structure of the device is as follows Figure 4 As shown, the device includes a memory 41 for storing computer-readable instructions and a processor 42 for executing the computer-readable instructions, wherein when the computer-readable instructions are executed by the processor, the processor is triggered to execute the method described.

[0107] The methods and / or embodiments in the embodiments of the present application may be implemented as computer software programs. For example, the embodiments of the present disclosure include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes program code for executing the method shown in the flowchart. When the computer program is executed by the processing unit, the above functions defined in the method of the present application are executed.

[0108] It should be noted that the computer-readable medium described in the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device.

[0109] In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0110] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0111] The flow chart or block diagram in the accompanying drawings shows the possible architecture, function and operation of the equipment, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated system for hardware that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0112] As another aspect, the embodiments of the present application further provide a computer-readable medium, which may be included in the device described in the above embodiments; or may exist independently without being assembled into the device. The above computer-readable medium carries one or more computer-readable instructions, which may be executed by a processor to implement the steps of the methods and / or technical solutions of the above multiple embodiments of the present application.

[0113] In a typical configuration of the present application, the terminal and the equipment of the service network each include one or more processors (CPU), input / output interface, network interface and memory.

[0114] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0115] Computer readable media include permanent and non-permanent, removable and non-removable media, and can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, modules of programs or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0116] In addition, an embodiment of the present application further provides a computer program, which is stored in a computer device, so that the computer device executes the method for controlling code execution.

[0117] It should be noted that the present application can be implemented in software and / or a combination of software and hardware, for example, can be implemented using an application specific integrated circuit (ASIC), a general purpose computer or any other similar hardware device. In certain embodiments, the software program of the present application can be executed by a processor to implement the above steps or functions. Similarly, the software program of the present application (including related data structures) can be stored in a computer-readable recording medium, for example, a RAM memory, a magnetic or optical drive or a floppy disk and similar devices. In addition, some steps or functions of the present application can be implemented using hardware, for example, as a circuit that cooperates with a processor to perform each step or function.

[0118] It is obvious to those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and that the present application can be implemented in other specific forms without departing from the spirit or basic features of the present application. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-restrictive, and the scope of the present application is limited by the attached claims rather than the above description, so it is intended to include all changes that fall within the meaning and scope of the equivalent elements of the claims in the present application. Any figure mark in the claims should not be regarded as limiting the claims involved. In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the device claim can also be implemented by one unit or device through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.

Claims

1. A defect detection method, characterized in that: include: Acquire an image to be detected of the object to be detected, input the image to be detected into a pre-trained defect detection model, and obtain defect detection categories output by the defect detection model and category confidences corresponding to each defect detection category; Determining candidate defect categories according to each of the defect detection categories and the category confidence corresponding to each of the defect detection categories; Obtaining a wafer feature map, and determining a wafer defect category based on the wafer feature map; Determine a target defect category of the image to be inspected according to the candidate defect categories and the wafer defect category; Wherein, determining the wafer defect category based on the wafer feature map includes: determining the wafer problem area according to the defect point distribution in the wafer feature map, and determining the wafer defect category based on the standard point distribution of the pre-constructed standard defect category and the defect point distribution; Wherein, when the number of the wafer problem areas is at least two, determining the target defect category of the image to be detected according to the candidate defect category and the wafer defect category includes: Determine the target defect category according to the wafer problem area, the wafer defect category and the candidate defect category; Wherein, determining the target defect category according to the wafer problem area, the wafer defect category and the candidate defect category includes: If the wafer defect category corresponds to the wafer problem area, and there is a category consistent with the wafer defect category in the candidate defect categories, then the wafer defect category is used as the target defect category; If the wafer defect category does not correspond to the wafer problem area and there is no defect category corresponding to the wafer problem area in the candidate defect categories, or the wafer defect category corresponds to the wafer problem area and there is a defect category corresponding to the wafer problem area in the candidate defect categories, then the category confidence of the candidate defect category is updated based on the position information of the candidate defect category, and the candidate defect category with the highest updated category confidence is used as the target defect category; If the wafer defect category does not correspond to the wafer problem area, and there is no defect category corresponding to the wafer problem area in the candidate defect categories, the candidate defect category with the highest category confidence is used as the target defect category.

2. The method according to claim 1, characterized in that The step of determining a candidate defect category according to each defect detection category and a category confidence level corresponding to each defect detection category includes: sorting the defect detection categories in reverse order based on the category confidence; Traversing is performed according to the reverse sorting order, and for the traversed current defect detection category, the candidate defect category is determined according to the category attribute information of the current defect detection category and the category confidence of the current defect detection category, until the traversal end condition is met.

3. The method according to claim 2, characterized in that The traversal end condition includes at least one of the following: The relationship between the category confidence of the current defect detection category and the category confidence of the next defect detection category does not satisfy the set condition; The number of candidate defect categories meets the set threshold.

4. The method according to claim 1, characterized in that: When the number of the wafer problem area is one, determining the target defect category of the image to be inspected according to the candidate defect category and the wafer defect category includes: If there is a category in the candidate defect categories that is consistent with the wafer defect category, the wafer defect category is used as the target defect category; If there is no category consistent with the wafer defect category among the candidate defect categories, the category with the highest category confidence among the candidate defect categories is used as the target defect category.

5. A defect detection device, characterized in that: include: The detection model prediction module is used to obtain an image to be detected of the object to be detected, input the image to be detected into a pre-trained defect detection model, and obtain the defect detection category output by the defect detection model and the category confidence corresponding to each defect detection category; A candidate category determination module, used to determine a candidate defect category according to each defect detection category and a category confidence corresponding to each defect detection category; A wafer defect classification module, used to obtain a wafer feature map and determine a wafer defect category based on the wafer feature map; A target category determination module, used to determine a target defect category of the image to be detected according to the candidate defect category and the wafer defect category; The wafer defect category module is used to determine the wafer problem area according to the defect point distribution in the wafer feature map, and determine the wafer defect category based on the standard point distribution of the pre-constructed standard defect category and the defect point distribution; Wherein, when the number of the wafer problem areas is at least two, the target category determination module is used to determine the target defect category according to the wafer problem areas, the wafer defect category and the candidate defect category; Wherein, the target category determination module is also used for: If the wafer defect category corresponds to the wafer problem area, and there is a category consistent with the wafer defect category in the candidate defect categories, then the wafer defect category is used as the target defect category; If the wafer defect category does not correspond to the wafer problem area and there is no defect category corresponding to the wafer problem area in the candidate defect categories, or the wafer defect category corresponds to the wafer problem area and there is a defect category corresponding to the wafer problem area in the candidate defect categories, then the category confidence of the candidate defect category is updated based on the position information of the candidate defect category, and the candidate defect category with the highest updated category confidence is used as the target defect category; If the wafer defect category does not correspond to the wafer problem area, and there is no defect category corresponding to the wafer problem area in the candidate defect categories, the candidate defect category with the highest category confidence is used as the target defect category.

6. A computer device, characterized in that: The device comprises: one or more processors; and A memory storing computer program instructions, which, when executed, cause the processor to perform the method according to any one of claims 1 to 4.

7. A computer-readable medium having computer program instructions stored thereon, wherein the computer program instructions can be executed by a processor to implement the method according to any one of claims 1 to 4.

Images