Inspection device, method for inspection, and program

Abstract

To detect a failure of attaching a joint member to a storage body, which is difficult to determine from the appearance.SOLUTION: An inspection device 1 includes: an image data acquisition unit 141 for acquiring monochrome taken image data generated by imaging an inspection target object W by a monochrome imaging device 2 with infrared light applied on the inspection target object W, the inspection target object W having a storage body A and a joint member B attached to the storage body A by a retainer C to cover an end part of the storage body A; and a failure detection unit 142 for detecting a failure of the inspection target object W on the basis of a state of at least one of the joint member B and the retainer C in the monochrome taken image data.SELECTED DRAWING: Figure 4

Description

[Technical field]

[0001] The present invention relates to an inspection device, an inspection method, and a program for inspecting an object to be inspected. [Background technology]

[0002] 2. Description of the Related Art Conventionally, there is known an inspection method for determining whether or not a container is sealed based on an image of the container captured by a camera (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2002-116152 A Summary of the Invention [Problem to be solved by the invention]

[0004] When sealing a housing by placing a connecting member over the housing and fixing it in place, even if the connecting member appears to be in close contact with the housing from the outside, the connecting member may not be in close contact with the housing. Conventional inspection methods have been unable to detect such defects.

[0005] Therefore, the present invention has been made in consideration of these points, and has an object to make it possible to detect improper attachment of a connecting member to a container, which is difficult to determine by appearance. [Means for solving the problem]

[0006] The inspection apparatus of the first aspect of the present invention comprises an image data acquisition unit that acquires monochrome captured image data generated by photographing an object to be inspected with a monochrome imaging device while irradiating infrared light on the object to be inspected, the object having a container and a connecting member attached to the container with a fastener so as to cover an end of the container, and a defect detection unit that detects defects in the object to be inspected based on the state of at least one of the connecting member and the fastener in the monochrome captured image data.

[0007] The defect detection unit may detect the defect that the container and the connecting member are not connected by the fastener in a state where they overlap by a predetermined amount or more.

[0008] The defect detection unit may detect the defect that the fastener does not have a predetermined shape.

[0009] The image data acquisition unit may further acquire color image data generated by photographing the object to be inspected with a color imaging device, and the defect detection unit may detect a first type of defect based on the monochrome image data and detect a second type of defect different from the first type based on the color image data.

[0010] The image data acquisition unit may acquire a plurality of monochrome captured image data sets generated by photographing the object to be inspected while it is transported through an inspection target range in an imaging area of ​​the monochrome imaging device, and the defect detection unit may detect defects in the object to be inspected based on the plurality of monochrome captured image data sets, each of which captures a partial area of ​​the object to be inspected.

[0011] The multiple monochrome captured image data sets are generated by photographing multiple different areas of the connecting member in the longitudinal direction of the connecting member, and the defect detection unit may determine that the inspected object is defective when it detects an abnormality in the condition of the fastener in combined image data obtained by combining the multiple monochrome captured image data sets.

[0012] The multiple monochrome captured image data sets are generated by a single imaging device taking images at different times while the object to be inspected is transported in the longitudinal direction of the connecting member, and the defect detection unit may detect defects in the object to be inspected based on the multiple monochrome captured image data sets after moving at least one of the monochrome captured image data sets in a direction perpendicular to the longitudinal direction of the connecting member so as to increase continuity of images of the connecting member or the fastener in two adjacent monochrome captured image data sets among the multiple monochrome captured image data sets.

[0013] The plurality of monochrome captured image data are generated by photographing a plurality of different areas of the connecting member, and the defect detection unit is a machine learning model trained by machine learning using teacher image data generated by photographing different areas of the connecting member attached to the container in a state where a defect is present, or teacher image data generated by photographing different areas of the connecting member attached to the container in a state where there is no defect, and defects in the object to be inspected may be detected by inputting the plurality of monochrome captured image data into a machine learning model that outputs a result of determining whether or not the object to be inspected has a defect.

[0014] An inspection method of a second aspect of the present invention is executed by a computer and includes the steps of: acquiring monochrome image data of an object to be inspected, the object having a container and a connecting member attached to the container with a fastener so as to cover an end of the container, by photographing the object to be inspected with a monochrome imaging device while irradiating the object with infrared light; and detecting defects in the object to be inspected based on the state of at least one of the connecting member and the fastener in the monochrome image data.

[0015] A third aspect of the program of the present invention is a program for causing a computer to execute the steps of: acquiring monochrome image data of an inspection object having a container and a connecting member attached to the container with a fastener so as to cover an end of the container, by photographing the inspection object with a monochrome imaging device while irradiating the inspection object with infrared light; and detecting defects in the inspection object based on the state of at least one of the connecting member and the fastener in the monochrome image data. Effect of the Invention

[0016] According to the present invention, it is possible to detect improper attachment of a connecting member to a container, which is difficult to determine by appearance alone. [Brief description of the drawings]

[0017] [Figure 1] FIG. 2 is a diagram for explaining an overview of an inspection system S. [Diagram 2] 11A and 11B are diagrams for explaining internal defects that can be inspected using monochrome image data. [Diagram 3] 11A and 11B are diagrams for explaining internal defects that can be inspected using monochrome image data. [Figure 4] 1 is a diagram showing a configuration of an inspection device 1. FIG. [Diagram 5] 4 is a diagram showing an example of test result data stored in the storage unit 13. FIG. [Figure 6] 13 is a diagram for explaining a process in which the defect detection section 142 detects defects in the inspection object W based on the combined image data. FIG. [Figure 7] FIG. 13 is a diagram showing partial image data captured in an inspection target range at a plurality of different times. [Figure 8] FIG. 13 is a diagram for explaining a method for creating combined image data. [Figure 9] 4 is a flowchart showing a process flow in the inspection system S. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] [Overview of Inspection System S] Fig. 1 is a diagram for explaining an overview of an inspection system S. The inspection system S is a system for inspecting an object W to be inspected, which has a container A and a connecting member B attached to the container A with a fastener C so as to cover an end of the container A. The container A is an object such as a bag or a box having a space for containing some kind of item. The container A is, for example, a bag that contains food such as rice or flour.

[0019] Although details will be described later, the connecting member B is, for example, a lid having a shape that covers the opening of the container A. The fastener C is a member for connecting the container A and the connecting member B by penetrating the container A and the connecting member B with a partial area of ​​the container A and a partial area of ​​the connecting member B overlapping each other, and is, for example, a thread or a linear metal (for example, a stapler core).

[0020] The inspection system S includes an inspection device 1, a monochrome imaging device 2, a color imaging device 3, and a conveying device 4. The inspection device 1 inspects the object W based on image data generated by the monochrome imaging device 2 and the color imaging device 3 by capturing images of the object W. The conveying device 4 is a device, such as a belt conveyer, for conveying the object W so that the object W passes through the imaging areas of the monochrome imaging device 2 and the color imaging device 3.

[0021] The monochrome imaging device 2 has a monochrome imaging element. In the inspection system S, the monochrome imaging device 2 generates monochrome captured image data (hereinafter, may be referred to as "monochrome image data") by capturing an image of the inspection object W while the inspection object W is irradiated with infrared light. The monochrome imaging device 2 may be an infrared imaging device (infrared camera) having a function of irradiating infrared light. The color imaging device 3 has a color imaging element and generates color captured image data (hereinafter, may be referred to as "color image data"). The monochrome imaging device 2 inputs the monochrome captured image data to the inspection device 1, and the color imaging device 3 inputs the color image data to the inspection device 1.

[0022] The items that the inspection device 1 can inspect based on the color image data are different from the items that the inspection device 1 can inspect based on the monochrome image data. Specifically, the inspection device 1 can inspect the inspection object W for defects in appearance, such as stains or tears, using the color image data. Meanwhile, since the monochrome imaging device 2 can sense infrared light, the inspection device 1 can inspect the inspection object W for internal defects that are not visible in appearance, such as an excessively small amount of overlap between the container A and the connecting member B, using the monochrome image data.

[0023] 2 and 3 are diagrams for explaining internal defects that can be inspected using monochrome image data. FIG. 2 shows a state in which a coupling member B is normally attached to a housing A. FIG. 2(a) is a front view of an object W to be inspected, and FIG. 2(b) is a cross-sectional view of the object W to be inspected along the line XX. As shown in FIG. 2(b), the coupling member B is formed in a mountain shape, and is coupled to the housing A in a state in which it overlaps the housing A so that its ridge line almost coincides with the end of the housing A. In the object W to be inspected shown in FIG. 2, the area of ​​the region in which the housing A and the coupling member B overlap is sufficiently large, so that the housing A and the coupling member B are coupled to each other by a fastener C, and the housing A is sealed.

[0024] In contrast, Fig. 3 shows a state in which the coupling member B is not properly attached to the housing A. In the inspection object W shown in Fig. 3, the area of ​​the region where the housing A and the coupling member B overlap is small, the housing A and the coupling member B are not coupled by the fastener C, and the housing A is not completely sealed. The inspection device 1 can detect such internal defects.

[0025] [Configuration of inspection device 1] 4 is a diagram showing the configuration of the inspection device 1. The inspection device 1 has a communication unit 11, a display unit 12, a storage unit 13, and a control unit 14. The control unit 14 has an image data acquisition unit 141 and a defect detection unit 142.

[0026] The communication unit 11 has a communication interface for transmitting and receiving data to and from an external device. For example, the communication unit 11 receives monochrome image data from the monochrome imaging device 2 and receives color image data from the color imaging device 3. The communication unit 11 inputs the received monochrome image data and color image data to the image data acquisition unit 141. The communication unit 11 may transmit the result of the defect detection unit 142 determining whether or not there is a defect to an external computer.

[0027] The display unit 12 is, for example, a display, and displays the result of the defect detection unit 142 determining whether or not there is a defect.

[0028] The storage unit 13 includes storage media such as a Read Only Memory (ROM), a Random Access Memory (RAM), and a Solid State Drive (SSD), etc. The storage unit 13 stores a program executed by the control unit 14.

[0029] The storage unit 13 stores various data necessary for inspecting the inspection object W. The storage unit 13 stores, for example, monochrome image data and color image data received by the communication unit 11. The storage unit 13 may store the monochrome image data and color image data in association with identification information for identifying the inspection object W.

[0030] The storage unit 13 also stores reference data for determining whether the inspection object W is good or bad. The storage unit 13 stores the reference data in association with each inspection item. The reference data may be data showing an image of a good product or data showing an image of a defective product. The reference data may be a machine learning model created by deep learning to distinguish between good and defective products.

[0031] The storage unit 13 also stores inspection result data indicating the result of the determination of the presence or absence of a defect by the defect detection unit 142. FIG.

[0032] 5, the judgment results of each inspection item based on the monochrome image data and the color image data are associated with an overall judgment result in association with the inspection object ID of the inspection object W. If all of the inspection items are pass (◯), the overall judgment result is "pass," and if at least one of the inspection items is fail (×), the overall judgment result is "fail."

[0033] "Shallow seam", an inspection item corresponding to monochrome image data, is a state in which the container A and the connecting member B overlap by more than a specified amount but are not connected by the fastener C, as shown in Figure 3. "Skipped stitch", an inspection item corresponding to color image data, is a state in which the thread is not sewn at a specified interval. "Oil stain" is a state in which oil or other stains are attached. "Tear" is a state in which part of the inspected object W is torn.

[0034] The control unit 14 includes, for example, a CPU (Central Processing Unit). The control unit 14 executes a program stored in the storage unit 13, thereby functioning as an image data acquisition unit 141 and a defect detection unit 142.

[0035] The image data acquisition unit 141 acquires, via the communication unit 11, monochrome image data generated by photographing the inspection object W with the monochrome imaging device 2 while irradiating the inspection object W with infrared light. The infrared light may be irradiated onto the inspection object W by the monochrome imaging device 2, or may be irradiated by another device.

[0036] The image data acquisition unit 141 acquires monochrome image data generated by capturing an image of the inspection object W irradiated with infrared light while the inspection object W is being transported so as to pass through the imaging area of ​​the monochrome imaging device 2. The image data acquisition unit 141 may further acquire color image data generated by capturing an image of the inspection object W by the color imaging device 3.

[0037] The image data acquisition unit 141 stores the acquired monochrome image data and color image data in the storage unit 13 in association with the inspection object ID. The image data acquisition unit 141 accepts the setting of the inspection object ID of the first inspection object W at the time when the inspection is started, for example, via the communication unit 11, and updates the inspection object ID according to a predetermined rule every time new monochrome image data and color image data are acquired, thereby determining the inspection object ID corresponding to each monochrome image data and color image data. When the inspection object W includes an inspection object ID or an image including an inspection object ID, the image data acquisition unit 141 may identify the inspection object ID by reading the inspection object ID from the monochrome image data or color image data.

[0038] The image data acquisition unit 141 may store the monochrome image data and color image data in the storage unit 13 in association with the time when the monochrome image data or color image data was captured together with the inspection object ID or as the inspection object ID.

[0039] The defect detection unit 142 detects defects of the inspection object W based on at least one of the monochrome image data and the color image data. The defect detection unit 142 detects a first type of defect based on the monochrome captured image data, and detects a second type of defect different from the first type based on the color captured image data. The first type of defect is, for example, a shallow seam that does not appear in the appearance of the inspection object W or a shape defect of the fastener C.

[0040] The defect detection unit 142 detects defects in the inspection object W based on, for example, the state of at least one of the connecting member B and the fastener C in the monochrome image data. As an example, the defect detection unit 142 detects a defect in which the container A and the connecting member B are not connected by the fastener C with an overlap of a predetermined amount or more (i.e., a "shallow stitch" defect). The predetermined amount is, for example, the length or area required to sew together the container A and the connecting member B with thread as the fastener C.

[0041] Specifically, the defect detection unit 142 determines that the product is a good product when the distance from the top end of the container A to the bottom end of the connecting member B is longer than the distance from the top end of the container A to the position of the thread. The defect detection unit 142 determines that the product is defective when the distance from the top end of the container A to the bottom end of the connecting member B is equal to or less than the distance from the top end of the container A to the position of the thread, or equal to or less than the distance from the top end of the container A to the position of the thread plus a margin. Although the top end of the container A cannot be seen from the outside when the connecting member B covers the container A, the defect detection unit 142 can detect defects that are not visible from the outside by using monochrome image data generated by the monochrome imaging device 2 that can detect infrared light.

[0042] The defect detection unit 142 may detect a defect that the fastener C does not have a predetermined shape. For example, the defect detection unit 142 reads out reference image data showing the shape of the fastener C in a normal state from the storage unit 13, and compares the shape of the fastener C shown in the monochrome image data with the shape of the fastener C shown in the reference image data. The defect detection unit 142 determines that the inspection object W is defective when there is a difference of a predetermined amount or more between the comparison results. A defect such as the fastener C being cut inside the joining member B does not appear on the outside, but by using the monochrome image data generated by the monochrome imaging device 2, the defect detection unit 142 can detect such a defect of the fastener C that does not appear on the outside.

[0043] [Inspection using multiple captured image data] However, if the object W to be inspected has width and it is difficult to illuminate it with a uniform illuminance, the illuminance varies depending on the part of the object W to be inspected, and therefore the defect detection accuracy may be reduced with only one piece of monochrome image data and one piece of color image data generated by the monochrome imaging device 2 or the color imaging device 3. Therefore, the image data acquisition unit 141 may acquire multiple pieces of monochrome image data generated by capturing images of different areas of the object W multiple times while the object W passes through the imaging area of ​​the monochrome imaging device 2.

[0044] The multiple monochrome image data are generated by one monochrome imaging device 2 capturing images of the inspection object W at different times while the inspection object W is being transported in the longitudinal direction of the joining member B, and each of the monochrome image data includes image data of multiple different regions on the joining member B. Similarly, the image data acquisition unit 141 may acquire multiple color image data generated by capturing images of different regions of the inspection object W multiple times while the inspection object W passes through the imaging region of the color imaging device 3.

[0045] The defect detection unit 142 detects defects in the object W based on at least one of a plurality of monochrome image data and a plurality of color image data, each of which shows a partial area of ​​the object W. However, if the defect detection unit 142 detects defects using each monochrome image data or color image data individually, there is a risk that defects existing near the ends of each monochrome image data or color image data may not be detected. Therefore, the defect detection unit 142 may determine that the object W is defective when it detects an abnormality in the state of the fastener C in the monochrome combined image data obtained by combining a plurality of monochrome image data. The defect detection unit 142 may determine that the object W is defective when it detects an abnormality such as oil staining in the color combined image data obtained by combining a plurality of color image data.

[0046] 6 to 8 are diagrams for explaining the process in which the defect detection unit 142 detects defects in the inspection object W based on the combined image data. In the example shown in FIG. 6, the inspection object W is transported so as to pass through the inspection target range in the order of times T1, T2, T3, and T4. The inspection target range includes, for example, the center of the imaging region of the monochrome imaging device 2, and is a range in which the inspection object W is best imaged. A bonding auxiliary line L is drawn in the horizontal direction on the inspection object W shown in FIG. 6. The inspection target range may be a range irradiated with infrared light.

[0047] Fig. 7 is a diagram showing partial image data captured in an inspection range at multiple different times (times T1, T2, T3, and T4). Fig. 8 is a diagram for explaining a method of creating combined image data by combining the multiple partial image data shown in Fig. 7.

[0048] The defect detection unit 142 creates tentative combined image data as shown in Fig. 8(a) by arranging the multiple partial image data in the order of the time when they were photographed. In the tentative combined image data shown in Fig. 8(a), the joining auxiliary line L is discontinuous because the inspection object W may have a curved or uneven shape, and the multiple partial image data are misaligned in the vertical direction due to vibrations while the inspection object W is being transported. If the defect detection unit 142 determines the presence or absence of a defect in this state, the determination accuracy will decrease.

[0049] Therefore, the defect detection unit 142 creates combined image data by correcting the tentative combined image data by moving at least one partial image data in a direction perpendicular to the longitudinal direction of the connecting member B so as to increase the continuity of the image of the connecting member B or the fastener C in two adjacent partial image data out of the plurality of partial image data (e.g., a plurality of monochrome partial image data). The defect detection unit 142 corrects the tentative combined image data by, for example, moving the positions of the plurality of partial image data so that the shape of the connection auxiliary line L in the tentative combined image data becomes a predetermined shape (e.g., a straight line). The defect detection unit 142 may correct pixel values ​​so that the boundary line between the plurality of partial image data becomes less noticeable.

[0050] The defect detection unit 142 detects defects in the inspection object W based on the corrected combined image data as shown in Fig. 8(b). The defect detection unit 142 detects defects in the inspection object W based on the combined image data corrected in this manner, thereby improving the accuracy of defect detection.

[0051] [Detecting defects using machine learning models] The defect detection unit 142 may use a machine learning model to detect defects in the inspection object W. The machine learning model is created by machine learning using, for example, teacher image data generated by photographing different areas of the coupling member B attached to the container A in a state where there is a defect, or teacher image data generated by photographing different areas of the coupling member B attached to the container A in a state where there is no defect. The machine learning model receives image data and outputs a result of determining whether or not the inspection object W has a defect.

[0052] The defect detection unit 142 detects defects in the inspection object W based on the results output from the machine learning model by inputting a plurality of monochrome partial image data or a plurality of color partial image data to the machine learning model. If the machine learning model has previously learned using teacher image data composed of a plurality of partial image data, the defect detection unit 142 can determine whether the inspection object W is good or bad based on the plurality of partial image data without combining the plurality of partial image data to create combined image data, thereby reducing the processing time.

[0053] [Processing flow in inspection system S] Fig. 9 is a flowchart showing the flow of processing in the inspection system S. The flowchart shown in Fig. 9 starts from the point when the inspection system S starts inspecting the object W to be inspected.

[0054] The monochrome imaging device 2 and the color imaging device 3 capture images of the inspection object W while the inspection object W is being transported (S11). The monochrome imaging device 2 and the color imaging device 3 capture images of the inspection object W, for example, based on the control of the inspection device 1. The image data acquisition unit 141 acquires a plurality of partial image data (S12), and the defect detection unit 142 combines the plurality of partial image data to create provisional combined image data (S13). The defect detection unit 142 corrects the provisional combined image data to create monochrome combined image data and color combined image data (S14).

[0055] The defect detection unit 142 analyzes the monochrome combined image data and the color combined image data to determine the quality of the inspection object W. The defect detection unit 142 may analyze the monochrome combined image data of the inspection object W, and if no defect is detected, may analyze the color combined image data of the inspection object W, or may analyze both the monochrome combined image data and the color combined image data. The following description will explain the process when the defect detection unit 142 analyzes both the monochrome combined image data and the color combined image data.

[0056] The defect detection unit 142 analyzes the monochrome image data (S15) and determines whether or not the lid as the joining member B overlaps with the container A by a predetermined amount or more (S16). If the defect detection unit 142 determines that the lid overlaps with the container A by a predetermined amount or more (YES in S16), it determines whether or not the shape of the fastener C is normal (S17). If the defect detection unit 142 determines that the shape of the fastener C is normal (YES in S17), it determines that the result of the inspection based on the monochrome image data is normal.

[0057] The defect detection unit 142 analyzes the color image data (S18) and judges whether or not there are skipped stitches in the fastener C (S19), whether or not there is oil staining (S20), and whether or not there are tears (S21). If all of these are normal (YES in S21), the defect detection unit 142 judges that the result of the inspection based on the color image data is normal. If both the result of the inspection based on the monochrome image data and the result of the inspection based on the color image data are normal, the defect detection unit 142 judges that the inspected object W is normal (S22).

[0058] If the defect detection unit 142 detects an abnormality in any of S16, S17, S19, S20, and S21, it determines that the inspection object W is abnormal (defective) (S23). The defect detection unit 142 repeats the processes from S11 to S23 until an operation to end the inspection is performed (NO in S24).

[0059] [Effects of Inspection Device 1] As described above, the image data acquisition unit 141 acquires monochrome image data generated by photographing the object W by the monochrome imaging device 2 while irradiating the object W with infrared light, and the defect detection unit 142 detects defects in the object W based on the state of at least one of the connecting members B and the fasteners C in the monochrome image data. The defect detection unit 142 analyzes the monochrome image data of the object W irradiated with infrared light in this manner, thereby making it possible to detect with high accuracy defects in the object W caused by abnormal states of the connecting members B or the fasteners C that are not apparent from the outside.

[0060] In the above description, the container A is a bag, the connecting member B is a lid that covers the container A, and the fastener C is a thread, but the types of the container A, the connecting member B, and the fastener C are arbitrary. As another example, the container A may be a cardboard box, the connecting member B may be a flat cardboard member, and the fastener C may be a metal member.

[0061] Although the present invention has been described above using the embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments, and various modifications and changes are possible within the scope of the gist of the present invention. For example, all or part of the device can be configured by distributing or integrating functionally or physically in any unit. In addition, new embodiments resulting from any combination of multiple embodiments are also included in the embodiments of the present invention. The effect of the new embodiment resulting from the combination combines the effect of the original embodiment. [Explanation of symbols]

[0062] 1. Inspection equipment 2. Monochrome imaging device 3. Color imaging device 4. Conveyor 11 Communications Department 12 Display section 13 Storage section 14 Control section 141 Image data acquisition unit 142 Defect detection unit

Claims

1. an image data acquisition unit that acquires monochrome captured image data generated by photographing an object to be inspected using a monochrome imaging device while irradiating the object with infrared light, the object having a container and a connecting member attached to the container with a fastener so as to cover an end of the container; a defect detection unit that detects defects in the inspection object based on a state of at least one of the connecting member and the fastener in the monochrome captured image data; An inspection device having the above structure.

2. The defect detection unit detects the defect that the container and the coupling member are not coupled by the fastener in a state where they overlap with each other by a predetermined amount or more.

2. The inspection device according to claim 1.

3. The defect detection unit detects the defect that the fastener does not have a predetermined shape.

2. The inspection device according to claim 1.

4. The image data acquisition unit further acquires color image data generated by photographing the object to be inspected using a color imaging device; the defect detection unit detects a first type of defect based on the monochrome captured image data, and detects a second type of defect different from the first type based on the color captured image data.

2. The inspection device according to claim 1.

5. the image data acquisition unit acquires a plurality of monochrome captured image data generated by capturing images of the inspection object while the inspection object is being transported through an inspection target area in an imaging region of the monochrome imaging device; the defect detection unit detects defects of the object to be inspected based on the plurality of monochrome captured image data, each of which includes a partial area of ​​the object to be inspected. The inspection device according to any one of claims 1 to 3.

6. the plurality of monochrome captured image data are generated by capturing images of a plurality of different regions of the connecting member in a longitudinal direction of the connecting member, the defect detection unit determines that the inspection object is defective when detecting an abnormality in a state of the fastener in combined image data obtained by combining the plurality of monochrome captured image data. The inspection device according to any one of claims 1 to 3.

7. the plurality of monochrome captured image data are generated by capturing images of the inspection object at different timings by a single imaging device while the inspection object is being transported in the longitudinal direction of the connecting member, the defect detection unit detects defects in the inspection object based on the plurality of monochrome captured image data after moving at least one of the monochrome captured image data in a direction perpendicular to the longitudinal direction of the connecting member so that continuity of images of the connecting member or the fastener in two adjacent monochrome captured image data pieces among the plurality of monochrome captured image data is increased.

7. The inspection apparatus according to claim 6.

8. the plurality of monochrome captured image data are generated by capturing images of a plurality of different regions of the connecting member, The defect detection unit is a machine learning model that is machine-learned using teacher image data generated by photographing different areas of the connecting member attached to the container when a defect is present, or teacher image data generated by photographing different areas of the connecting member attached to the container when no defect is present, and detects defects in the object to be inspected by inputting the plurality of monochrome captured image data into the machine learning model that outputs a result of determining whether or not the object to be inspected has a defect.

7. The inspection apparatus according to claim 6.

9. The computer executes a step of acquiring monochrome captured image data generated by photographing an object to be inspected, the object having a container and a connecting member attached to the container with a fastener so as to cover an end of the container, with a monochrome imaging device while irradiating the object with infrared light; detecting a defect of the inspection object based on a state of at least one of the connecting member and the fastener in the monochrome captured image data; An inspection method having the following features.

10. On the computer, a step of acquiring monochrome captured image data generated by photographing an object to be inspected, the object having a container and a connecting member attached to the container with a fastener so as to cover an end of the container, with a monochrome imaging device while irradiating the object with infrared light; detecting a defect of the inspection object based on a state of at least one of the connecting member and the fastener in the monochrome captured image data; A program for executing the above.

Images