Inspection system, management device, inspection method, program, recording medium, and article manufacturing method

Abstract

To alter the inspection condition of a preprocess inspection step on the basis of fault results in a post-process inspection step and improve the inspection accuracy of the preprocess inspection step.SOLUTION: The present inspection system comprises an imaging unit for imaging a workpiece, and a processing unit for assessing the acceptability of the workpiece on the basis of the image having been imaged by the imaging unit. When it is assessed in a second inspection step on the basis of a second image imaged by the imaging unit that the workpiece has a defect, the processing unit extracts a candidate for the defect portion of the workpiece on the basis of the second image and a first image having been imaged by the imaging unit in a first inspection step that precedes the second inspection step. Furthermore, the inspection condition of the first inspection step is altered on the basis of the candidate for the defect portion of the workpiece.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present invention relates to an inspection system, a management device, an inspection method, a program, a recording medium, and a method for manufacturing an article. [Background technology]

[0002] Users' demands for product quality are increasing year by year, and so are the demands placed on the appearance of the products. For this reason, manufacturers are being asked to visualize and evaluate subtle differences in appearance, such as minute scratches, dirt, stains, coating defects, and uneven coloring of the film, during appearance inspections before the products are shipped.

[0003] Patent Document 1 discloses an inspection method and an inspection device that performs image processing on an image including an object to determine the presence or absence of a defect. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2014-66599 A Summary of the Invention [Problem to be solved by the invention]

[0005] However, there are multiple inspection processes before a product is shipped, and there are cases where a visual defect that was not apparent in the upstream inspection process is discovered for the first time in the downstream inspection process. In this case, the processing cost up to the downstream process is wasted, and additional costs (rework) are incurred by returning to the upstream process and redoing the work. [Means for solving the problem]

[0006] The inspection system for solving the above problem includes an imaging unit that images a workpiece, and Including the workA processing unit that judges whether the workpiece is good or bad based on the image, wherein the processing unit Including the workpiece When it is determined that the workpiece is defective based on the second image, the second image and a first inspection process performed before the second inspection process are Including the workpiece Based on the first image, tree, The inspection conditions of the first inspection step are changed. Effect of the Invention

[0007] According to the present invention, since the inspection conditions of the front-end inspection process are changed based on the defective result of the back-end inspection process, the inspection accuracy of the front-end inspection process can be improved and the processing costs and redoing costs incurred up to the back-end inspection process can be reduced. [Brief description of the drawings]

[0008] [Figure 1] FIG. 1 is a schematic diagram showing an embodiment of an inspection system according to a first embodiment. [Diagram 2] FIG. 2 is a schematic diagram showing an embodiment of a processing section according to the first embodiment. [Diagram 3] FIG. 11 is a flow chart showing an embodiment of an inspection method according to the second embodiment. [Figure 4] FIG. 11 is a flow chart showing an embodiment of an inspection method according to the third embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] (Inspection System) [First embodiment] Fig. 1 is a schematic diagram of an inspection system 1 according to the first embodiment. The inspection system 1 includes an inspection device 300 and a processing device (processing unit) 100. Fig. 2 is a block diagram of the processing device 100.

[0010] The inspection device 300 is a device that inspects the workpiece 206. The inspection device 300 includes an imaging unit 304, an illumination unit 305, and a base 307.

[0011] The imaging unit 304 has an imaging body 3041 having an imaging element (not shown) in a housing, a lens unit 3042 having an optical system, and an imaging drive unit 3043 that moves the positions of the imaging body 3041 and the lens unit 3042. A commercially available digital camera can be used for the imaging body 3041 and the lens unit 3042. The imaging unit 304 can be moved in the Z direction and the XY direction by the control unit 105 provided in the processing device 100. Therefore, the relative position between the imaging unit 304 and the workpiece 206 can be changed arbitrarily. By changing this relative position, the imaging area of ​​the workpiece 206 and the focal position with respect to the workpiece 206 can be changed.

[0012] The illumination unit 305 has at least one light source and has a function of irradiating light onto the surface of the workpiece 206, which is the inspection target of the inspection device 300. The light source is, for example, an LED. The illumination unit 305 is attached to a support (not shown) and can be moved in the Z direction and the XY direction by the control unit 105 provided in the processing device 100. Therefore, the relative position between the illumination unit 305 and the workpiece 206 can be changed arbitrarily. In addition, the control unit 105 can control ON / OFF of the power supply of the light source of the illumination unit 305. The number and shape of the light sources of the illumination unit 305 are selected according to the type of the workpiece 206. For example, a ring shape is suitable for a workpiece that is circular when viewed in a plane from the Z direction, such as an optical lens. In addition, for a workpiece with a large shape, such as the exterior of a car, it is suitable to use multiple light sources and arrange them so that they can be irradiated in sequence from multiple directions. The wavelength of the light source is selected according to the type of the workpiece 206 and the type of appearance defect to be detected. For example, for three-dimensional defects with fine irregularities such as scratches or peeling of thin films, blue light with a wavelength in the range of 420 nm to 500 nm is preferable. Furthermore, a high S / N ratio is preferable to determine the quality of the appearance. It is also preferable to arrange the LEDs at a high density and use LEDs that emit linear light.

[0013] The base 307 is a member on which the workpiece 206 to be inspected is placed. In the embodiment of FIG. 1, the base 307 does not have a driving mechanism capable of moving in the Z direction or the XY direction, but a driving mechanism may be provided on the base 307, and the relative positions of the workpiece 206 and the illumination unit 305 or the imaging unit 304 may be controlled using the control unit 105. Also, in FIG. 1, the workpiece 206 is an optical lens, and is a concave lens. However, the shape of the workpiece 206 is not limited to this, and may be a flat plate or a convex shape. Also, the workpiece may be a non-light-transmitting workpiece or a light-transmitting workpiece.

[0014] In FIG. 1, the inspection device 300 is in a state in which one inspection device can communicate with one processing device 100, but the number of inspection devices 300 is not limited to one. One processing device 100 may be capable of communicating with multiple inspection devices 300. In addition, in multiple inspection systems 1, each inspection system may be controlled using a central processing device that centrally controls the processing devices 100 of each inspection system 1. The central processing device may control each processing device through an external server, and may be located in a remote location away from the location of the inspection device 300. In addition, a commercially available three-dimensional shape measuring device may be used as the inspection device 300.

[0015] The processing device 100 has a function of judging the quality of the workpiece 206 based on an image captured by the imaging unit 304. The processing device 100 also has a function of managing the inspection conditions of the inspection device 300. The processing device 100 has an information storage unit 101, an image processing unit 102, an estimation unit 103, an output unit 104, a control unit 105, and an input unit 106.

[0016] The information storage unit 101 has a function of storing information specific to the workpiece 206 and inspection conditions for the inspection process. The information specific to the workpiece 206 is, for example, the shape, material, and part number. In addition, the result of pass / fail judgment, processing data for the workpiece 206, and the like may be associated with these pieces of information. The inspection conditions include conditions for image processing performed by the image processing unit 102 on the image, and conditions for judging the pass / fail of the workpiece based on the image (pass / fail judgment criteria). In addition to the above conditions, the inspection conditions may include the position of the imaging unit, the aperture and focal length of the imaging unit, the exposure conditions of the imaging unit, the relative position of the imaging unit and the workpiece, the position of the illumination unit, the wavelength of the illumination light of the illumination unit, the relative position of the illumination unit and the workpiece, and the like. The information storage unit 101 stores data to be transmitted to the image processing unit 102 and the estimation unit 103, and outputs these data to the image processing unit 102 and the estimation unit 103. The information storage unit 101 may store not only information in the inspection device 300, but also information via a network from a cloud in which a database is stored or an external device.

[0017] The image processing unit 102 has a function of reading out the image processing conditions stored in the information storage unit 101, processing the image stored in the information storage unit 101, and outputting the image or image information after the image processing to the estimation unit 103. Here, the image processing can be appropriately selected from known processing methods such as image conversion, deformation processing, and feature extraction. In this case, a machine learning function related to the image processing method may be incorporated inside the image processing unit 102, and thus only the image processing method suitable for estimation may be selectively output. Note that the image captured by the imaging unit 304 does not necessarily need to be processed by the image processing unit 102, and an image without processing may be output to the estimation unit 103.

[0018] The estimation unit 103 judges the quality of the workpiece 206 using the image and image information output from the image processing unit 102 and various information stored in the information storage unit 101. The conditions for judging the quality of the workpiece 206 from the image include, for example, the minimum size at which a part such as a scratch, dust, stain, coating defect, or color unevenness of the film is treated as a defective part, and the position at which the defective part occurs. The minimum size to be treated as a defective part may be set according to the position of the workpiece 206 to be inspected. Also, if there is a processing step before the inspection step, the size and position of the defective part may be set according to the processing step. The estimation unit 103 judges the quality of the workpiece 206 depending on whether these conditions are satisfied.

[0019] The estimation unit 103 also has a function of estimating the location of a defect in the workpiece 206, using the image and image information output from the image processing unit 102 and various information stored in the information storage unit 101. The procedure for estimating the location of a defect will be explained in the section on the inspection method.

[0020] The estimation unit 103 may be provided with a machine learning unit for machine learning, and a computer equipped with multiple types of GPUs (Graphics Processing Units) may be used for the calculation. Known methods can be used for machine learning, and specifically, they can be broadly divided into "supervised learning", "unsupervised learning", and "reinforcement learning".

[0021] When "supervised learning" is used, the input data includes images of the workpiece in the previous process (the first inspection process described below), information about the workpiece, information about the processing process, image processing conditions applied to the images in the previous process, etc. The teacher data includes images of the workpiece that was determined to be defective in the subsequent process (the second inspection process described below), information about the workpiece, information about the processing process, and the determination results. A learning model is constructed by estimating the similarity between the defective parts of the workpiece and the defect type estimated in the previous process from the input data and the teacher data, and outputting each of them. A parameter weighting calculation can be performed to estimate the similarity, and the weighting can be changed.

[0022] When "unsupervised learning" is used, there are images of a workpiece that has been determined to be defective in a later process, images of the workpiece in a previous process, and conditions for image processing to be applied to each of these images.

[0023] The output unit 104 has a function of outputting the image captured by the imaging unit 304 and the image processed by the image processing unit 102 to the display device 202. In addition, the output unit 104 can also output information stored in the information storage unit 101 to the display device 202.

[0024] The control unit 105 is connected to the information storage unit 101, the image processing unit 102, the estimation unit 103, and the output unit 104, and sets various conditions and the number of times estimation is performed. The control unit 105 can also control the driving of the inspection device 300 and the inspection conditions of the inspection device 300. The control unit 105 also has a function of changing the inspection conditions after the estimation unit 103 estimates the defective part, and storing the changed inspection conditions in the information storage unit 101.

[0025] The input unit 106 has a function of temporarily storing information such as the shape, material, and part number of the workpiece 206 input by a user via a user interface device 201 described later. The input unit 106 is, for example, a RAM or a HDD. The input unit 106 does not necessarily need to be connected to the UI device 201, and may be connected to an external device (not shown) such as a CAD, and information on the workpiece 206 may be input to the input unit 106 from the external device. Also, information on the workpiece 206 may be input to the input unit 106 from a database stored in a cloud via a network (not shown).

[0026] The processing device 100 is connected to a user interface device 201 (hereinafter referred to as UI device 201 ) and a display device 202 .

[0027] 1, the UI device 201 is connected to the input unit 106. The UI device 201 is, for example, an input device such as a keyboard, a mouse, or a touch pen.

[0028] The display device 202 is connected to the output unit 104 of the processing device 100 in Fig. 1. The display device 202 is a display, and is preferably capable of displaying images with a high gray scale, such as a high dynamic range display.

[0029] (Testing method) [Second embodiment] Next, a description will be given of a method for inspecting the workpiece 206 using the inspection system 1. Fig. 3 is a flow chart showing an embodiment of the inspection method according to the second embodiment.

[0030] First, the workpiece 206 is placed on the base 307 of the inspection system 1, and a first image including the surface of the workpiece 206 opposite to the side in contact with the base 307 is acquired (imaged) by the imaging unit 304 (S11). The acquired first image is stored in the information storage unit 101 in association with the inspection conditions of the first image, information specific to the workpiece 206, the process history performed up to that point, and the like. For example, the information specific to the workpiece 206 can be selected by the user via the UI device 201 from information stored in the information storage unit 101. The inspection conditions include conditions for image processing performed by the image processing unit 102 on the image, and conditions for determining the quality of the workpiece based on the image. In addition to the above conditions, the inspection conditions may include imaging conditions such as the position of the imaging unit, the aperture and focal length of the imaging unit, the exposure conditions of the imaging unit, the relative position of the imaging unit and the workpiece, the position of the illumination unit, the wavelength of the illumination light of the illumination unit, and the relative position of the illumination unit and the workpiece. Note that the first image does not necessarily need to be image-processed.

[0031] Next, the estimation unit 103 of the processing device 100 judges whether or not a defect has occurred on the surface of the workpiece 206 based on the acquired first image (S12). The steps from S11 to S12 are called the first inspection process. The first image and the judgment result of the estimation unit 103 are displayed on the display device 202 via the output unit 104, and are visible to the user. If it is judged that the workpiece 206 has a defect from the first image acquired in the first inspection process, the first inspection process is repeated, or the process does not proceed to the process after the first inspection process.

[0032] Next, the workpiece 206 that has been determined to have no defects by the first inspection process is processed (S13). The type of processing performed on the workpiece is not particularly limited, but may be, for example, a film-forming process such as deposition, spray coating, spin coating, squeegeeing, and sputtering. In addition to film-forming, processes such as cutting, coating, molding, grinding, polishing, drilling, and deformation processing may also be used. By performing such processing, it becomes more likely that an appearance defect that was not apparent in the first inspection process will be discovered for the first time in the subsequent second inspection process. However, it is not necessary to have a processing process between the first inspection process and the second inspection process described later.

[0033] Next, the second inspection process (S21, S22) is performed on the processed workpiece 206. In the second inspection process, first, a second image including the same surface as the surface of the workpiece 206 from which the first image was obtained is obtained (imaged) by the imaging unit 304 (S21). The obtained second image is stored in the information storage unit 101 in association with the inspection conditions of the second image, information specific to the workpiece 206, the process history performed up to that point, and the like. The type of information associated here may be the same as or different from that of the first image. Note that there may be a plurality of imaging units 304, and the first imaging unit performing the first inspection process and the second imaging unit performing the second inspection process may be different devices. The second image is displayed on the display device 202 via the output unit 104, and is in a state where it can be viewed by the user.

[0034] Next, based on the acquired second image, the estimation unit 103 of the processing device 100 judges whether or not a defect occurs on the surface of the workpiece 206 (S22). Specifically, the estimation unit 103 judges whether or not the workpiece is good or bad based on the image and image information output from the image processing unit 102 and the pass / fail judgment criteria based on various information stored in the information storage unit 101. The pass / fail judgment criteria for the image of the workpiece include, for example, the minimum size at which a part such as a scratch, dust, stain, coating defect, or color unevenness of the film is treated as a defective part, and the position at which the defective part occurs. The minimum size to be treated as a defective part may be set according to the position of the workpiece to be inspected. In addition, if there is a processing step before the inspection step, the size and position of the defective part may be set according to the processing step. The estimation unit 103 judges whether or not the workpiece 206 is good or bad depending on whether or not these conditions are satisfied. The result of the pass / fail judgment is displayed on the display device 202 via the output unit 104, and is in a state where the user can see it.

[0035] If it is determined that there is no defect (N in S22), the inspection ends. On the other hand, if it is determined that a defect has occurred in the workpiece 206 (Y in S22), that is, if a defect that was not found in the first inspection process is found in the workpiece 206 in the second inspection process, the process proceeds to the next step (S23).

[0036] When it is determined that the workpiece 206 is defective, the control unit 105 reads out information on the defective determination in the second inspection process from the information storage unit 101 (S23). Specifically, information on the workpiece that has been determined to be defective is read out, and the read out information is provided to the estimation unit 103 in association with the second image (S23). Examples of information to be associated with the second image include the material, shape, and size of the workpiece, a machining drawing used in the machining process, machining process information, a determination criterion for determining that the workpiece is defective, and a defect classification. In addition, part number information, manufacturing information, design data, and the like of the workpiece may also be included.

[0037] Next, based on the read information, the control unit 105 reads out the first image from the information storage unit 101 and provides the read out first image to the estimation unit 103 (S24). If there are multiple inspection processes before the second inspection process, the first image may be read out from any of the multiple inspection processes performed before the second inspection process, but it is preferable to read out information from the immediately preceding inspection process. When reading out the first image, the estimation unit 103 may also read out information about the workpiece and the inspection conditions when the first image was acquired.

[0038] Next, the estimation unit 103 extracts information on the defective portion from the second image, and estimates the defective portion of the workpiece based on the second image and the first image (S25). Specifically, a candidate defective portion is extracted from the second image, and a defective portion candidate area where the defective portion exists is estimated. Based on the estimated defective portion candidate area, the type and position of the image processing to be performed on the first image are determined. Then, a defective portion where the defect occurs is extracted by the image processing, and the position of the defective portion is specified in the first image. There may be one or more defective portion candidates and defective portion candidate areas. For the image processing, a general image processing method related to image data, which performs image conversion, deformation, and extraction of information such as feature amount, can be used. Among them, contrast adjustment, noise removal, binarization processing, pattern recognition, and expansion / contraction image processing are preferable, and by combining these, defective portions that did not meet the defect judgment standard in the first inspection process can be highlighted and recognized as defective portions. By combining and comparing the defective parts in the second image, the defect candidate areas in the first image, and the processing result of the first image, the accuracy of estimating the defective parts in the first image can be improved. In addition, one of the reasons why the defective parts become apparent in the second process is the positional shift or rotation of the workpiece when it is imaged. Therefore, it is desirable for the estimation unit 103 to estimate the defective parts after aligning the positional relationship of the entire workpiece in the first image and the second image.

[0039] Furthermore, the estimation unit 103 can use information on the processing step in S12 when setting the defect occurrence candidate region. For example, if the processing step is a film formation step, and the film formation means is deposition, spray coating, or the like, in which the film is formed from the top of the inspection surface, the defect occurrence candidate region in the first image is often almost the same as the defect occurrence location in the second image. However, if the film is formed by coating, such as spin coating or squeegee, the defective portion is often located closer to the coating start point than the defect occurrence region. In this way, by reflecting the type of processing step in the setting (condition) of the defect occurrence candidate region, the accuracy of estimating the defect location in the first image can be increased.

[0040] Next, the estimation unit 103 estimates the type of defect (defect classification) at the defective portion. Types of defects include scratches, cloudiness, coating defects, uneven color of the film, dirt, and dust. The user lists these types of defects in advance and stores them in the information storage unit 101 in association with an image including the workpiece, whereby the estimation unit 103 estimates the type of defect. The estimation unit 103 also estimates the size of the defect at the defective portion. The size of the defect can be estimated based on the information on the size of the defect extracted from the first image and the information on the size of the defect extracted from the defect in the second image.

[0041] Then, based on the image processing conditions and the information on the defective part when the defective part in the first image was estimated, new inspection conditions for the first inspection process are created, and the inspection process for the first inspection process stored in the information storage unit 101 is changed and registered (S26). Items to be changed as inspection conditions include the image processing conditions for the first image, and pass / fail judgment conditions for recognizing the defect in the first image as a defective part. In addition to the above conditions, imaging conditions such as the position of the imaging unit, the aperture and focal length of the imaging unit, the exposure conditions of the imaging unit, the relative position of the imaging unit and the workpiece, the position of the illumination unit, the wavelength of the illumination light of the illumination unit, and the relative position of the illumination unit and the workpiece may be included.

[0042] The flow shown in Figure 3 may be executed at any time for all workpieces determined to be defective, or may be executed only for workpiece numbers specified by the user, or may be executed only for defective part numbers that have been detected a certain number of times or more, and the user may also specify the frequency of execution.

[0043] As described above, according to the inspection method of the second embodiment, the inspection conditions of the first inspection process are changed based on the defective result of the second inspection process, so that the inspection accuracy of the first inspection process can be improved. Also, by changing the inspection conditions of the first inspection process, it becomes possible to estimate the pass / fail judgment of the second inspection process after the processing process performed after the first inspection process from the result of the first inspection process.

[0044] [Third embodiment] Fig. 4 is a flow diagram showing one implementation of the inspection method according to the third embodiment. The third embodiment differs from the flow of the inspection method of the second embodiment in the method of estimating defective parts (S25), and can improve the estimation accuracy more than the second embodiment. Since S11 to S24 are the same as the flow of Fig. 3, S11, S12, S13, S21, S22, S23, and S24 are omitted in Fig. 4.

[0045] An arbitrary image processing method is selected for the first image captured in the first inspection process read out in S24, and image processing is performed (S251). The result is compared with the defective judgment work information, and the defect occurrence location is estimated in the first image (S252). If there is no candidate for a defective part that is estimated to be the cause of the defect, it is determined whether or not there are other image processing conditions for the first image (S253). If there are other image processing conditions, the process returns to S251, and if not, the flow ends.

[0046] If there is a candidate for a defect that may be a defective part based on the estimation, the image processing conditions and judgment conditions selected in S251 are added as provisional inspection conditions for the first process (S254, S255). Note that the order of these two processes may be reversed.

[0047] Next, the number of inspections specified by the device or user is performed in the first inspection process (S256), and it is confirmed whether the judgment rate specified by the device or user can be obtained (S257). At this time, the inspection objects for which the specified number of inspections are performed may be workpieces that have already been inspected, or new workpieces. The judgment rate is the ratio (proportion) of the number of defectives in the first inspection process to the number of defectives in the second inspection process, as judged under the provisional inspection conditions of the first process. If this judgment rate is equal to or higher than the specified judgment rate, it can be determined that the inspection conditions of the first inspection process are highly accurate.

[0048] The image processing conditions (S261) for the first inspection process determined by the above flow and the judgment criteria conditions (S262) at that time are registered, and the flow ends. At this time, there is no problem even if the order of S261 and S262 is reversed. In addition, by repeating the flow shown in Figure 4 for the same type of defect, the accuracy of the judgment criteria is improved.

[0049] (Production method of the article) [Fourth embodiment] The inspection system according to the above-described embodiment can also be used in a manufacturing method of an article. The manufacturing method of an article according to the fourth embodiment includes a step of inspecting the surface of a workpiece using the above-described inspection system, and a step of performing a predetermined process on the inspected workpiece. The predetermined process is, for example, at least one selected from processing, conveying, and assembly. Here, processing includes film formation, cutting, coating, painting, molding, grinding, polishing, drilling, and deformation processing. The manufacturing method of an article according to this embodiment is advantageous in terms of processing costs compared to conventional methods.

[0050] In the above embodiment, an optical lens is used as an example of the workpiece, but the present embodiment can be applied to other transparent workpieces such as optical lenses. It can be not only transparent but also non-transparent, and can be applied to the inspection of the surfaces of automobile bodies and jewels.

[0051] In addition, a program for realizing one or more functions of the above-mentioned embodiments can be provided to a system or device via a network or a recording medium, and one or more processors in the computer of the system or device can read and execute the program. Also, the program can be realized by a circuit (e.g., ASIC) that realizes one or more functions. [Explanation of symbols]

[0052] 100 Processing device (processing unit) 101 Information storage section 102 Image Processing Department 103 Estimation part 104 Output section 105 Control section 106 Input section 201 UI device 202 Display device 206 Work 300 Inspection Equipment 304 Imaging unit 3041 Imaging body 3042 Lens section 3043 Imaging drive unit 305 Lighting Department 307 Foundation

Claims

1. An imaging unit that images a workpiece; A processing unit that judges whether the workpiece is good or bad based on an image including the workpiece captured by the imaging unit, The processing unit includes: When it is determined that the work is defective based on a second image including the work in the second inspection process, The inspection conditions of the first inspection process are changed based on the second image and a first image including the workpiece in a first inspection process performed before the second inspection process.

1. An inspection system comprising:

2. The processing unit includes: Extracting candidates for defective portions of the workpiece based on the second image and the first image; 2. The inspection system according to claim 1, further comprising: changing an inspection condition for the first inspection step based on the extracted candidate defective portions of the workpiece.

3. The inspection system according to claim 2 , wherein the processing unit estimates a defective portion of the workpiece by performing image processing on the first image.

4. The inspection system according to claim 1 , wherein the inspection conditions of the first inspection process that are changed include conditions of image processing of the first image.

5. The inspection system according to claim 1 , wherein the inspection conditions of the first inspection process that are changed include imaging conditions of the imaging unit.

6. The inspection system according to any one of claims 2 to 5, wherein the processing unit judges the quality of the work based on a pre-constructed learning model, and reflects changes in the inspection conditions of the first inspection process based on extracted candidates for defective parts of the work in the learning model.

7. 7. The inspection system according to claim 1, further comprising a processing step for processing the workpiece, the processing step being carried out between the first inspection step and the second inspection step.

8. The inspection system according to claim 7 , wherein the processing unit is capable of estimating a pass / fail judgment of the workpiece after the machining process in the first inspection process by changing an inspection condition of the first inspection process.

9. the imaging unit includes a first imaging unit and a second imaging unit, The first imaging unit images the workpiece in the first inspection process, The inspection system according to claim 1 , wherein the second imaging unit images the workpiece in the second inspection process.

10. An inspection system as described in any one of claims 1 to 9, wherein when the processing unit determines that the workpiece is not defective based on the first image including the workpiece in the first inspection process, and when the processing unit determines that the workpiece is defective based on the second image including the workpiece in the second inspection process, the processing unit changes the inspection conditions for the first inspection process based on the second image and the first image including the workpiece in a first inspection process performed prior to the second inspection process.

11. A management device that manages an inspection condition of a first inspection process and an inspection condition of a second inspection process that is performed after the first inspection process, The management device includes: When it is determined that the workpiece is defective based on a second image including the workpiece used for inspection in the second inspection step, The inspection conditions of the first inspection process are changed based on the second image and a first image including the workpiece used for inspection in the first inspection process. A management device comprising:

12. A first inspection step of capturing a first image including the workpiece; A second inspection step of capturing a second image including the workpiece after the first inspection step and determining whether the workpiece is good or bad based on the second image, When the processing unit determines that the workpiece is defective in the second inspection process, a step of changing an inspection condition of the first inspection step based on the second image and the first image by the processing unit; An inspection method comprising:

13. A program for causing a computer to execute each step of the inspection method according to claim 12.

14. A computer-readable recording medium having the program according to claim 13 recorded thereon.

15. Inspecting a workpiece using the inspection system according to any one of claims 1 to 10; A step of performing a predetermined process on the inspected workpiece; A method for producing an article, comprising:

Images