TW202238455A - Training data creation assistance apparatus and training data creation assistance method - Google Patents

Abstract

The present invention relates to a training data creation assistance device, wherein abnormal data representing an image of an inspection target object determined to be defective in advance is acquired by an abnormal data acquisition unit. Normal data representing an image of the normal inspection target object is acquired by a normal data acquisition unit in a manner corresponding to the abnormal data acquired by the abnormal data acquisition unit. Difference data representing a difference between the abnormal data acquired by the abnormal data acquisition unit and the normal data acquired by the normal data acquisition unit in a manner corresponding to the abnormal data is generated by a difference data generation unit; Based on the difference data generated by the difference data generation unit, a presentation unit presents an image of a portion of the inspection target object.

Description

Training data production support device and training data production support method

The present invention relates to a training data production support device and a training data production support method for supporting training data production.

Products such as products and foods are properly inspected to determine whether they are good or not in each step from the manufacturing process to distribution. For example, Japanese Patent Laid-Open No. 2011-119471 discloses a defect inspection device for inspecting various defects generated during the manufacturing steps of semiconductor wafers.

In this defect inspection apparatus, a plurality of SEM (scanning electron microscope) images showing a plurality of wafers to be inspected are generated. The user designates a SEM image showing a good wafer as a template from among the plurality of generated SEM images. By calculating the difference between each of a plurality of SEM images (inspection images) other than the template and the template, it is used to detect defects on the circuit pattern of the wafer shown in the inspection image.

In the defect inspection device, there are cases where it may be judged to have a defect even if it is an inspection target object that does not have a defect. In this case, although it does not have a defect, the inspection target object is discarded due to an erroneous inspection result being obtained, resulting in a decrease in yield. Therefore, it is desired to perform inspection with higher accuracy.

An object of the present invention is to provide a training data creation support device and a training data creation support method capable of easily creating training data that can be inspected with high precision.

(1) A training data creation support device according to an aspect of the present invention, which supports the creation of training data used for re-inspection of inspection objects, and is equipped with: an abnormal data acquisition unit, which acquires abnormal data, and the abnormal data is Displaying an image of an inspection object judged to have a defect in advance; a normal data acquisition unit that acquires an image showing a normal inspection object in a manner corresponding to the abnormal data acquired by the abnormal data acquisition unit Normal data; a differential data generation unit that generates differential data showing the abnormal data obtained by the abnormal data acquisition unit and the normal data obtained by the normal data acquisition unit in a manner corresponding to the abnormal data. a data difference; and a presentation unit for presenting an image of a portion of the inspection object based on the difference data generated by the difference data generation unit. In this training material creation support device, the inspection is presented based on the difference data showing the difference between the abnormal data showing the image of the inspection object judged to be defective in advance and the normal data showing the image of the normal inspection object. An image of a part of an object. There is a high possibility that re-inspection is required for the part of the inspection target object in the image displayed based on the difference data. Therefore, the user can create training materials for rechecking the part by confirming the displayed image. With this, it is possible to easily create training data for inspection using high precision.

(2) The training data production support device may be further equipped with: a reception unit that accepts the selection of differential data presented by the presentation unit; and a registration unit that accepts the selected differential data for registration by the reception unit as training material. In this case, the difference data selected by the user is registered as the training data. In this way, it is easier to create training materials.

(3) The acceptance unit can further accept the correction of the difference data presented by the prompt unit, and the registration unit can also register the corrected difference data as training data. In this case, it is possible to more appropriately correct the portion of the image based on the difference data that requires re-inspection of the inspection object. With this, it can produce training data that uses higher precision for inspection.

(4) The normal data acquisition unit can also obtain a plurality of normal data in a manner corresponding to the abnormal data, and the differential data generation unit can also generate a plurality of differential data based on the abnormal data and the plurality of normal data corresponding to the abnormal data . In this case, a plurality of difference data are generated from one anomalous data. Thereby, the operation efficiency of training material production can be improved.

(5) The normal data acquisition part can also obtain a plurality of normal data corresponding to the abnormal data, and the difference data generation part can also generate a difference based on the average of the abnormal data and the plurality of normal data corresponding to the abnormal data material. According to this configuration, even if a noise component unrelated to a defect is accidentally mixed into any of the plurality of normal data, the plurality of normal data is averaged, so the noise component is hardly equal to the averaged normal data. The pixel value of the data is affected. Therefore, training data for inspection with higher precision can be produced.

(6) The normal data obtaining unit can also obtain normal data corresponding to a plurality of abnormal data, and the differential data generating unit can also generate differential data based on each abnormal data and the normal data corresponding to the abnormal data. In this case, training data can be created at high speed using common normal data.

(7) The normal data acquired by the normal data acquisition unit may include image data showing images of inspection objects that have not been previously judged to have defects. In this case, it is possible to easily obtain normal data showing an image of a normal inspection object.

(8) The normal data obtained by the normal data acquisition unit may also include master data, which is a design drawing showing the object to be inspected. In this case, it is possible to easily obtain normal data showing an image of a normal inspection object.

(9) The normal data acquisition department can also obtain the master data corrected according to the processing accuracy of the inspection object as normal data. According to this configuration, even when the inspection target area is fine, normal data showing an image of a normal inspection target object can be easily acquired.

(10) The abnormal data obtained by the abnormal data acquisition unit and the normal data obtained by the normal data acquisition unit are set as non-inspection target areas, and the differential data generation unit can also set the set non-inspection target areas Excluded to generate differential data, and so on. In this case, it is possible to prevent the image displayed by the differential data from including parts other than the inspection target area. Thereby, it is possible to create training data for performing higher-precision inspections.

(11) The abnormal data acquisition department can further obtain defect information, which shows the type of defect of the inspection object corresponding to the obtained abnormal data, and the differential data generation department can also provide the generated differential data. Defect information acquired by the Abnormal Data Acquisition Department. In this case, the user does not need to perform the operation of providing defect information to the training data. Thereby, the burden on the user can be reduced, and the work efficiency of training material production can be improved. In addition, since it does not cause errors accompanying the user's work, more accurate training materials can be produced.

(12) The abnormal data acquisition unit can also acquire binarized abnormal data, and the normal data acquisition unit can also acquire binarized normal data. In this case, since the amount of abnormal data and normal data is reduced, training data can be created at high speed.

(13) Another aspect of the present invention is a training data production support method, which supports the production of training data used for re-inspection of inspection objects, and includes the following steps: acquiring abnormal data Steps, the abnormal data is to display the image of the inspection object judged to have defects in advance; the step of obtaining normal data showing the image of the normal inspection object in a manner corresponding to the obtained abnormal data; generating The step of differential data, the differential data is to show the difference between the obtained abnormal data and the normal data obtained in a manner corresponding to the abnormal data; and a diagram showing the portion of the object to be inspected based on the generated differential data like steps.

When the support method is created based on the training data, it is presented based on the difference data showing the difference between the abnormal data showing the image of the inspection object judged to be defective in advance and the normal data showing the image of the normal inspection object. An image of a portion of an object to be inspected. The part of the inspection object in the image displayed based on the differential data is highly likely to require re-inspection. Therefore, the user can create training data for rechecking the part by confirming the displayed image. This makes it possible to easily create training data that can be inspected with higher accuracy.

[1] First Embodiment (1) Processing system Hereinafter, a training data production support device and a training data production support method according to an embodiment of the present invention will be described using drawings. In the following description, the training data creation support device is simply referred to as a support device. FIG. 1 is a diagram showing the configuration of a processing system including a supporting device according to a first embodiment of the present invention. As shown in FIG. 1 , the processing system 100 includes a processing device 10 , an inspection device 20 and a database storage device 30 .

The processing device 10 is composed of a CPU (Central Processing Unit) 11, a RAM (Random Access Memory) 12, a ROM (Read Only Memory) 13, a memory device 14, an operating unit 15, a display device 16 and an input/output I/F (Interface) 17 constitutes. The CPU 11 , RAM 12 , ROM 13 , memory device 14 , operation unit 15 , display device 16 , and input/output I/F 17 are connected to a bus bar 18 .

RAM12 is used as a working area of CPU11. There are system programs in ROM13 memory. The memory device 14 is a storage medium including a hard disk or a semiconductor memory, and a memory training data creation support program (hereinafter referred to as a support program for short). The supporting programs can also be stored in ROM 13 or other external memory devices. A support device 40 for executing training data creation support processing (hereinafter, simply referred to as support processing) is constituted by the CPU 11 , RAM 12 , and ROM 13 . Support and process the production of its support training materials.

The operation unit 15 is an input device such as a keyboard, a mouse, or a touch panel. The user can provide predetermined instructions to the support device 40 by operating the operation unit 15 . The display device 16 is a display device such as a liquid crystal display device, and displays a GUI (Graphical User Interface) or the like for accepting instructions from a user. The input/output I/F 17 is connected to the inspection device 20 .

The inspection device 20 is, for example, an AOI (Automatic Optical Inspection) device, and generates a plurality of image data showing images of a plurality of inspection objects by sequentially photographing the inspection objects, and stores the generated images. like data. A unique identification number is assigned to each stored image data.

Hereinafter, the inspection device 20 will be described by taking a substrate as an example of an inspection object, but the inspection object is not limited to the substrate. Furthermore, the so-called substrate refers to a substrate for FPD (Flat Panel Display) such as a semiconductor substrate, a liquid crystal display device, or an organic EL (Electro Luminescence) display device, a substrate for an optical disk, a substrate for a magnetic disk, a substrate for a magneto-optical disk, and a substrate for a photomask. Substrates, ceramic substrates, substrates for solar cells, etc.

The inspection device 20 processes the stored image data according to a predetermined algorithm, and inspects the substrate corresponding to each image data. The inspection device 20 may inspect the substrate corresponding to each image data according to the deep learning. During the inspection, it is determined whether the substrate has defects. In addition, for a substrate judged to have a defect, the type of the defect is judged.

In the inspection apparatus 20, even if it is a board|substrate which does not have a defect, it may be judged as having a defect. If there is no defect, but the substrate is discarded due to wrong determination, the yield rate will decrease. Therefore, the substrates judged to be defective are re-inspected by supervised learning. The support device 40 supports creation of training data used for rechecking. The database memory device 30 is a large-capacity memory device including a server or the like. In the database storage device 30, the training data that has been created is registered. Hereinafter, the detailed configuration of the support device 40 will be described.

(2) Support device FIG. 2 is a diagram showing the configuration of the supporting device 40 in FIG. 1 . FIG. 3 is a diagram showing various data used in creation of training data. FIG. 4 is a diagram showing an example of a display screen of the display device 16 during preparation of training materials. As shown in FIG. 2 , the support device 40 includes an abnormal data acquisition unit 41 , a normal data acquisition unit 42 , a differential data generation unit 43 , a presentation unit 44 , a reception unit 45 and a registration unit 46 as functional units. The CPU 11 in FIG. 1 executes the support program stored in the ROM 13 or the memory device 14, thereby realizing the functional part of the support device 40. Part or all of the functional parts of the support device 40 may also be realized by hardware such as electronic circuits.

The abnormality data acquiring unit 41 acquires image data (hereinafter referred to as abnormality data) from the inspection device 20 , and the image data is an image of each substrate judged to have a defect in advance by the inspection device 20 . The image data can display the image of the whole substrate, and if it is the same area, it can also display the image of a part of the substrate. A part of the image of the substrate based on the abnormality data acquired by the abnormality data acquisition part 41 is shown in the upper left part of FIG. 3 .

The normal data acquiring unit 42 acquires image data (hereinafter, referred to as normal data) showing a normal board image so as to correspond to each abnormal data acquired by the abnormal data acquiring unit 41 . In this example, the normal data is predetermined image data showing an image of a substrate not previously determined to have a defect by the inspection device 20 , and is acquired from the inspection device 20 . In the lower left part of FIG. 3 , a part of the image of the substrate based on the normal data obtained by the normal data obtaining part 42 is shown.

Abnormal data correspond to normal data. For example, in the case that abnormal data is obtained by the abnormal data obtaining unit 41, the normal data obtaining unit 42 may also obtain an image having an identification number preceding the identification number of the abnormal data and displaying a normal substrate image. data, as normal data. Or, when the abnormal data is obtained by the abnormal data acquisition part 41, the normal data acquisition part 42 can also obtain the image data having an identification number next to the identification number of the abnormal data and displaying a normal substrate image. , as normal data.

The differential data generation unit 43 calculates the difference between the abnormal data acquired by the abnormal data acquisition unit 41 and the normal data acquired by the normal data acquisition unit 42 in a manner corresponding to the abnormal data. , to generate new image data. Here, the image data generated by the difference data generation unit 43 is referred to as difference data. In the right part of FIG. 3 , a substrate image based on the differential data generated by the differential data generating unit 43 is displayed. The differential data is an image showing a portion of the board that is likely to require re-inspection. Thus, the differential data can be used as a marker showing which parts of the substrate need to be re-inspected.

The presentation unit 44 presents the difference data to the user by causing the display device 16 to display a GUI 50 ( FIG. 4 ) including an image based on the respective difference data generated by the difference data generation unit 43 . As shown in FIG. 4 , the GUI 50 includes an image display area 51 , a login key 52 , and a correction key 53 . A plurality of images of the object to be measured are displayed in the image display area 51 . In this example, the image based on the differential data is displayed in the image display area 51 in a manner of overlapping with the image based on the abnormal data, but it is also possible to display only the image based on the differential data in the image display area 51 .

The accepting unit 45 accepts the instruction to register the difference data. Specifically, the acceptance unit 45 accepts from the operation unit 15 the selection of the registered difference data as training data in the GUI 50 displayed by the presentation unit 44 . The user can select an arbitrary image by using the operation unit 15 while recognizing the image displayed in the image display area 51, and by operating the registration key 52, select the differential data displayed on the image as an instruction for training data. It is supplied to the receiving unit 45 . The registration unit 46 registers the difference data selected by the acceptance unit 45 in the database memory device 30 as training data.

In addition, the accepting unit 45 can accept correction of difference data. The user selects an arbitrary image in the image display area 51 by using the operation unit 15 , and instructs the receiving unit 45 to correct the difference data corresponding to the image by operating the correction key 53 . In addition, the user may use the operation unit 15 to perform coloring, etc. for specifying a portion to be re-inspected in the selected image.

When the designation is accepted by the accepting unit 45, the differential data generating unit 43 gives the selected differential data a mark indicating a portion where the designation is accepted. In this way, the differential data is corrected. After the difference data is corrected, when the registration key 52 is operated, the registration unit 46 registers the corrected difference data as training data in the database memory device 30 .

(3) Support processing FIG. 5 is a flowchart showing support processing performed by the support device 40 of FIG. 2 . The support processing in FIG. 5 is executed by CPU 11 in FIG. 1 executing a support program stored in ROM 13 or memory device 14 on RAM 12 . Hereinafter, support processing will be described using the support device 40 in FIG. 2 and the flowchart in FIG. 5 .

First, the abnormality data acquisition part 41 acquires each abnormality data from the inspection apparatus 20 (step S1). Next, the normal data acquisition part 42 acquires normal data corresponding to each abnormal data acquired in step S1 from the inspection apparatus 20 (step S2). Steps S1 and S2 can also be executed simultaneously.

Then, the difference data generation unit 43 generates each difference data based on the corresponding abnormal data and normal data respectively acquired in steps S1 and S2 (step S3). Then, the presentation unit 44 presents each differential data to the user by causing the display device 16 to display an image based on each differential data generated in step S3 (step S4).

Next, the acceptance unit 45 judges whether to accept correction of any one of the difference data presented in step S4 (step S5). When the correction of the difference data is not accepted, the accepting unit 45 proceeds to step S7. When the correction of any one of the differential data is accepted, the differential data generation unit 43 corrects the differential data by marking the differential data indicating a portion where a designation of correction is accepted (step S6), and proceeds to step S7.

In step S7, the acceptance unit 45 judges whether there is an instruction to register any one of the difference data presented in step S4 or the difference data corrected in step S6 (step S7). When there is no instruction to register the difference data, the accepting unit 45 proceeds to step S9. When registration of any one of the differential data is instructed, the registration unit 46 registers the instructed differential data as training data in the database memory device 30 (step S8), and proceeds to step S9.

In step S9, the registration unit 46 determines whether there is an instruction to end (step S9). The user can instruct to end or continue by performing a predetermined operation using the operation unit 15 . When there is no instruction to end, the registration unit 46 returns to step S5. When further registering differential data, the user instructs to continue without instructing to end. When the instruction ends, the registration unit 46 ends the support process.

(4) Effect In the support device 40 of this embodiment, the abnormality data displaying the image of the inspection object judged to have a defect in advance is acquired by the abnormality data acquisition part 41. The normal data showing the image of the normal inspection object is acquired by the normal data acquisition unit 42 in such a manner as to correspond to the abnormal data acquired by the abnormal data acquisition unit 41 . Differential data showing the difference between the abnormal data obtained by the abnormal data obtaining part 41 and the normal data obtained by the normal data obtaining part 42 in a manner corresponding to the abnormal data is generated by the differential data generating part 43. difference. In the image displayed based on the difference data, there is a high possibility that a part of the inspection object needs to be re-inspected.

Therefore, the partial image of the object to be inspected is presented by the presentation unit 44 based on the difference data generated by the difference data generation unit 43 . The accepting unit 45 accepts the selection of the difference data presented by the presenting unit 44 . The registration unit 46 registers the difference data selected by the accepting unit 45 in the database storage device 30 as training data. In this case, the user selects the differential data corresponding to the desired image while recognizing the suggested image, thereby registering the selected differential data as training data. In this way, it is easier to create training materials.

In addition, the accepting unit 45 further accepts correction of the difference data presented by the presenting unit 44 . The registration unit 46 registers the corrected difference data as training data. In this case, the user can more appropriately correct the portion of the image that needs to be re-inspected of the inspection object based on the difference data. Thereby, it is possible to create training data that can perform higher-precision inspection.

(5) Modification FIG. 6 is a diagram showing various data used for creation of training data in the first modified example. In the first modified example, the user may pre-register the non-inspection target area outside the displayed inspection target area in the inspection device 20 . When a non-inspection target area is registered, the inspection device 20 sets a non-inspection target area in the generated image data.

Therefore, as shown in the upper left part of FIG. 6 , a non-inspection target area is set in the abnormality data obtained by the abnormality data obtaining unit 41 . Similarly, as shown in the lower left part of FIG. 6 , a non-inspection target area is set in the normal data acquired by the normal data acquisition unit 42 . In this case, as shown in the right part of FIG. 6, when the set non-inspection object area is excluded, the differential data generation unit 43 calculates the difference between the abnormal data and the normal data corresponding to the abnormal data. Difference of pixel values to generate difference data. The pixel value of the non-inspection object area in the differential data can also be set to 0.

In this case, it is possible to prevent a mark showing a portion requiring re-inspection from being provided for a portion outside the inspection target area. This makes it possible to create training data for inspection with higher accuracy. In addition, since parts outside the inspection target area are not inspected, board re-inspection can be performed at high speed by using the prepared training data.

FIG. 7 is a diagram showing various data used for creation of training data in the second modification. As shown in the upper left part of FIG. 7 , the abnormality data acquiring unit 41 acquires abnormality data, and acquires defect information indicating the defect type of the substrate in the image displayed by the abnormality data. As shown in the right part of FIG. 7 , the differential data generating unit 43 provides the generated differential data with defect information acquired by the abnormal data acquiring unit 41 . In the image display area 51 of the GUI 50, based on the difference data, the image can also display the type of defect (such as color).

In this case, the user does not need to perform the operation of providing defect information to the training data. Thereby, the burden on the user can be reduced, and the work efficiency of training material production can be improved. In addition, more accurate training materials can be created because no errors will occur due to user-induced operations.

FIG. 8 is a diagram showing various data used for creation of training data in the third modified example. In the third modification, the user can preset the contents of binarization processing on the image data in the inspection device 20 . When it is set to perform binarization processing, the inspection device 20 generates binarized image data.

Therefore, as shown in the upper left part of FIG. 8 , the abnormality data acquisition unit 41 acquires the abnormality data subjected to the binarization process. Similarly, as shown in the lower left part of FIG. 8 , the normal data acquisition unit 42 acquires binarized normal data. As shown in the right part of FIG. 8 , the differential data generating unit 43 generates differential data by calculating the difference between the binarized abnormal data and the pixel values corresponding to the binarized normal data. .

In this case, since the amount of image data is reduced, training data can be created at high speed. In addition, by using the prepared training data, the re-inspection of the board can be performed at high speed.

[2] Second Embodiment In the first embodiment, although the normal data acquisition unit 42 acquires normal data in such a manner that one normal data corresponds to one abnormal data, this embodiment is not limited to this. The difference between the support processing in the second to fourth embodiments and the support processing in the first embodiment will be described below.

Fig. 9 is a diagram showing various data used for creation of training data in the second embodiment. In this embodiment, as shown in the lower left part of FIG. 9 , a plurality of normal data is acquired in such a manner that a plurality of normal data corresponds to one abnormal data. The differential data generation unit 43 generates differential data by calculating the difference between the pixel values of an abnormal data and normal data corresponding to the abnormal data. Therefore, as shown in the right part of FIG. 9 , a plurality of difference data are generated corresponding to one abnormal data.

According to this configuration, a plurality of difference data are generated from one abnormal data. Thereby, the operation efficiency of training material production can be improved.

[3] Third Embodiment Fig. 10 is a diagram showing various data used for creation of training data in the third embodiment. In this embodiment, as shown in the lower left part of FIG. 10, a plurality of normal data is obtained in such a manner that a plurality of normal data corresponds to one abnormal data. The differential data generation unit 43 generates differential data by calculating the difference between one abnormal data and the average pixel value of a plurality of normal data corresponding to the abnormal data. Therefore, as shown in the right part of FIG. 10 , one difference data is generated corresponding to one abnormal data.

According to this configuration, even if a noise component unrelated to a defect is accidentally mixed in any of the plurality of normal data, the plurality of normal data is averaged, so the noise component hardly affects the pixel value of the averaged normal data. make an impact. Therefore, by using the averaged normal data, it is possible to create training data for inspection with higher accuracy.

[4] Fourth Embodiment Fig. 11 is a diagram showing various data used for creation of training data in the fourth embodiment. In this embodiment, as shown in the upper left part of FIG. 11 , normal data is obtained in such a manner that one normal data corresponds to a plurality of abnormal data. The difference data generation unit 43 generates difference data based on each abnormal data and normal data corresponding to the abnormal data. Therefore, as shown in the right part of FIG. 11 , a plurality of difference data corresponding to a plurality of abnormal data are generated.

According to this configuration, since normal data does not need to be obtained every time abnormal data is obtained, training data can be created at high speed using common normal data. In this embodiment, image data showing an image of a normal substrate prepared in advance may be acquired by the normal data acquisition unit 42 as normal data. In this case, step S2 in the support process can also be executed before step S1.

The image data prepared in advance may be any of the image data generated by the inspection device 20, but may also be master data such as CAD data showing a design drawing of the substrate. Here, in the image data and master data generated by the inspection device 20, the pattern width of the substrate in the image or the radius of curvature of the corner of the pattern, etc., are affected by the processing of the substrate by etching or the like. Accuracy varies. Therefore, the master data can be corrected by changing the width of the pattern in the image or the radius of curvature of the corner of the pattern according to the processing accuracy of the substrate.

In addition, even in the first to third embodiments, although the image data showing the image of the substrate that was not previously judged to have a defect by the inspection device 20 was acquired from the inspection device 20 as normal data, the present embodiment Not limited to this. At least one of the normal data may be pre-prepared master data or corrected master data.

[5] Other embodiments In the above-mentioned embodiment, the support device 40 includes the receiving unit 45 and the registration unit 46, but this embodiment is not limited thereto. The support device 40 may not include the receiving unit 45 and the registration unit 46 . In this case, the user can also create training data for rechecking the part by confirming the image prompted by the GUI 50 . Thereby, it is easy to create training materials that can perform high-precision inspection.

[6] Correspondence between each constituent element of the claim and each part of the embodiment Hereinafter, although the corresponding example of each constituent requirement of a claim and each requirement of an embodiment is demonstrated, this invention is not limited to the following example. As each constituent element of the claim, various other elements having the constitution or function described in the claim may also be used.

In the above embodiment, the support device 40 is an example of the training data creation support device, the abnormal data acquisition unit 41 is an example of the abnormal data acquisition unit, and the normal data acquisition unit 42 is an example of the normal data acquisition unit. The differential data generation unit 43 is an example of a differential data generation unit, the presentation unit 44 is an example of a presentation unit, the reception unit 45 is an example of a reception unit, and the registration unit 46 is an example of a registration unit.

10: Processing device 11: CPU (central processing unit) 12: RAM (random access memory) 13: ROM (read-only memory) 14: memory device 15: Operation department 16: Display device 17: Input and output I/F (interface) 18: busbar 20: Check device 30: Database memory device 40: Support device 41: Abnormal Data Acquisition Department 42: Normal Data Acquisition Department 43: Differential data generation department 44: Prompt Department 45: Reception Department 46: Registration Department 50: GUI 51: Image display area 52: Login key 53: Correction key 100: Processing system

FIG. 1 is a diagram showing the configuration of a processing system including a support device according to a first embodiment of the present invention. FIG. 2 is a diagram showing the configuration of the supporting device in FIG. 1 . Fig. 3 is a diagram showing various kinds of data used in the creation of training data. Fig. 4 is a diagram showing an example of a display screen of a display device during preparation of training materials. FIG. 5 is a flowchart showing support processing performed by the support device in FIG. 2 . FIG. 6 is a diagram showing various data used for creation of training data in the first modification. FIG. 7 is a diagram showing various data used for creation of training data in the second modified example. FIG. 8 is a diagram showing various data used for creation of training data in the third modified example. Fig. 9 is a diagram showing various kinds of data used for creation of training data in the second embodiment. Fig. 10 is a diagram showing various data used for creation of training data in the third embodiment. Fig. 11 is a diagram showing various data used for creation of training data in the fourth embodiment.

15: Operation Department

16: Display device

20: Check device

30: Database memory device

40: Support device

41: Abnormal Data Acquisition Department

42: Normal Data Acquisition Department

43: Differential data generation department

44: Prompt Department

45: Reception Department

46: Registration Department

100: Processing system

Claims (13)

A training data production support device that supports the production of training data used for re-inspection of inspection objects; it has: An abnormality data acquisition unit, which acquires abnormality data, which is an image showing an inspection target object judged to have a defect in advance; A normal data acquisition unit, which acquires normal data showing images of normal inspection objects in a manner corresponding to the abnormal data acquired by the abnormal data acquisition unit; A differential data generation unit that generates differential data showing the abnormal data obtained by the abnormal data acquisition unit and the normal data obtained by the normal data acquisition unit in a manner corresponding to the abnormal data difference; and A presentation unit presents an image of a part of the inspection object based on the difference data generated by the difference data generation unit. For example, the training data production support device of claim item 1, which further includes: Acceptance section, which accepts the selection of differential data presented by the above-mentioned presentation section; and The registration unit registers the selected difference data as training data by the acceptance unit. The training data production support device according to Claim 2, wherein the acceptance unit further accepts correction of the difference data presented by the presentation unit, The registration unit registers the corrected difference data as training data. As for the training data production support device according to any one of claims 1 to 3, wherein the above-mentioned normal data acquisition part acquires a plurality of normal data in a manner corresponding to abnormal data, The above-mentioned difference data generation unit generates a plurality of difference data according to the abnormal data and the plurality of normal data corresponding to the abnormal data. As for the training data production support device according to any one of claims 1 to 3, wherein the above-mentioned normal data acquisition part acquires a plurality of normal data in a manner corresponding to abnormal data, The above-mentioned differential data generation unit generates differential data based on the average of the abnormal data and a plurality of normal data corresponding to the abnormal data. As for the training data production support device according to any one of claims 1 to 3, wherein the above-mentioned normal data acquisition part acquires normal data in a manner corresponding to a plurality of abnormal data, The differential data generation unit generates differential data based on each abnormal data and normal data corresponding to the abnormal data. The training data production support device according to any one of Claims 1 to 3, wherein the normal data acquired by the normal data acquisition unit includes an image displaying an inspection object that has not been judged to have a defect in advance Image data. The training data production support device according to any one of claims 1 to 3, wherein the normal data acquired by the normal data acquisition unit includes master data showing a design drawing of an object to be inspected. In the training data production support device according to claim 8, wherein the normal data obtaining unit obtains master data corrected according to the machining accuracy of the object to be inspected as the normal data. The training data production support device according to any one of claims 1 to 3, wherein non-inspection objects are set for the abnormal data obtained by the abnormal data acquisition part and the normal data obtained by the normal data acquisition part. area, The above-mentioned differential data generation unit excludes the set non-inspection target area to generate differential data. The training data production support device according to any one of claims 1 to 3, wherein the above-mentioned abnormal data obtaining unit further obtains defect information showing the inspection object corresponding to the obtained abnormal data. the type of defect, The above-mentioned difference data generation unit provides the generated difference data with the defect information obtained by the above-mentioned abnormal data acquisition unit. The training data production support device according to any one of claims 1 to 3, wherein the above-mentioned abnormal data acquisition unit acquires the binarized abnormal data, The above-mentioned normal data acquisition part acquires the binarized normal data. A training data production support method for supporting production of training data used for re-inspection of inspection objects, including the following steps: A step of obtaining abnormal data, which is an image showing an inspection target object judged to have a defect in advance; A step of obtaining normal data showing images of normal inspection objects in a manner corresponding to the obtained abnormal data; the step of generating differential data showing the difference between abnormal data obtained and normal data obtained in a manner corresponding to the abnormal data; and A step of presenting an image of a portion of an inspection object based on the generated difference data.

Images