Inspection apparatus

Abstract

To provide an inspection apparatus capable of simplifying the setting.SOLUTION: An X-ray inspection apparatus 10 comprises: a conveying unit 5 that conveys an article; an image generation unit 14 that images the article conveyed by the conveying unit 5 and generates an X-ray image; a storage unit 15 that stores a learned model generated using a set of the X-ray image generated by the image generation unit 14 and the weight or grade rank of the article shown in the X-ray image as teacher data; a control unit 16 that inputs the X-ray image of the article conveyed by the conveying unit 5 into the learned model, estimates the weight or the grade rank of the article, and generates a distribution signal on the basis of the estimated weight or grade rank of the article; and an output unit 17 that outputs the distribution signal to a distribution device 30 provided on the downstream side in the conveying direction of the article in the conveying unit 5.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to an inspection device. [Background technology]

[0002] Known conventional inspection devices include, for example, the device described in Patent Document 1. The inspection device described in Patent Document 1 includes an X-ray irradiation unit that irradiates an object with X-rays, an X-ray detection unit that detects the X-rays that have been irradiated from the X-ray irradiation unit and transmitted through the object, and a mass estimation unit that estimates the mass of the object based on the amount of X-rays detected by the X-ray detection unit. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2010-145135 Summary of the Invention [Problem to be solved by the invention]

[0004] In a configuration like the above inspection device that estimates the mass of an item based on the X-ray dose, it is necessary to adjust parameters to match the machine differences of the sensors that make up the X-ray detection unit and the characteristics of the item. Adjusting parameters requires the skill and effort of the operator, and as a result, setting up the inspection device was not easy.

[0005] An object of one aspect of the present invention is to provide an inspection device that can simplify settings. [Means for solving the problem]

[0006] (1) An inspection device according to one aspect of the present invention includes a conveying unit that conveys items; an image generating unit that photographs the items conveyed by the conveying unit and generates X-ray images; a memory unit that stores a trained model that is generated using pairs of the X-ray images generated by the image generating unit and the weight or grade rank of the items shown in the X-ray images as training data; a control unit that inputs the X-ray images of the items being conveyed by the conveying unit into the trained model, estimates the weight or grade rank of the items, and generates a sorting signal based on the estimated weight or grade rank of the items; and an output unit that outputs the sorting signal to a sorting device located downstream in the conveying unit in the direction in which the items are conveyed.

[0007] An inspection device according to one aspect of the present invention stores a trained model generated using training data that includes a pair of an X-ray image generated by an image generation unit and the weight or grade rank of an item shown in the X-ray image. The inspection device inputs an X-ray image of an item being transported by a transport unit into the trained model, and estimates the weight or grade rank of the item. In this way, the inspection device uses the trained model generated using the training data, thereby eliminating the need to adjust parameters to accommodate differences in sensors, etc., or the characteristics of the item. Therefore, the inspection device does not require an operator to adjust parameters. This simplifies the setup of the inspection device.

[0008] (2) In the inspection device of (1) above, the trained model may be generated using training data that is a pair of an X-ray image and a weight or grade based on a weighing signal output from a weighing device that detects the weight of an item. In this configuration, the trained model is generated using training data that includes the accurate weight or grade of an item actually weighed by the weighing device, thereby improving the estimation accuracy of the trained model. As a result, the inspection device can improve the inspection accuracy.

[0009] (3) In the inspection device described in (1) or (2) above, the memory unit may store a virtual X-ray image of a virtual item that is not the subject of inspection, and the control unit may input the virtual X-ray image into the trained model when the conveying unit starts conveying the item, and perform virtual processing to estimate the weight or grade rank of the virtual item. When the conveying unit starts conveying, such as when the inspection device is turned on, the control unit has not yet performed the estimation process, so the initial estimation process may take some time. This may result in a situation where the control unit is unable to generate a sorting signal by the time the item, whose weight or grade rank has been estimated, reaches the sorting device. Therefore, in the inspection device, the control unit inputs the virtual X-ray image into the trained model when the conveying unit starts conveying the item, and performs virtual processing to estimate the weight or grade rank of the virtual item. As a result, the inspection device can appropriately process the items that are actually the subject of inspection, because the inspection device has performed virtual processing once before actually inspecting the items.

[0010] (4) In any one of the inspection devices (1) to (3) above, if the control unit is unable to generate a sorting signal before the estimated weight or grade of an item reaches the sorting device, the control unit may cause the output unit to output a signal instructing the sorting device to eject the item. In this configuration, if the sorting signal cannot be generated in time, the item is forcibly ejected. This prevents defective items from being sent to a subsequent process. [Effects of the Invention]

[0011] According to one aspect of the present invention, it is possible to simplify the settings. [Brief explanation of the drawings]

[0012] [Figure 1] FIG. 1 is a diagram showing an X-ray inspection system including an X-ray inspection apparatus according to an embodiment. [Figure 2] FIG. 2 is a diagram showing an X-ray inspection device. [Figure 3] FIG. 3 is a diagram showing the internal configuration of the shielding box shown in FIG. [Figure 4] FIG. 4 is a diagram showing the configuration of the control unit. [Figure 5] FIG. 5 is a diagram illustrating a neural network. DETAILED DESCRIPTION OF THE INVENTION

[0013] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the description of the drawings, the same or corresponding elements are designated by the same reference numerals, and redundant description will be omitted.

[0014] (Configuration during X-ray inspection) The configuration for performing X-ray inspection will be described. As shown in FIG. 1, the X-ray inspection system 1 includes an X-ray inspection device 10, a sorting device 30, a weighing device 40, and a machine learning device 50. The X-ray inspection system 1 inspects an item A using a trained model generated by machine learning. However, as will be described later, the weighing device 40 only needs to be installed when generating the trained model.

[0015] The X-ray inspection device 10 includes a device main body 2, support legs 3, a shielding box 4, a transport unit 5, an X-ray irradiation unit 6, an X-ray detection unit 7, a display operation unit 8, and a control unit 9. The X-ray inspection device 10 generates an X-ray transmission image of the item A while transporting the item A, and inspects the item A (weight inspection, grade rank inspection) based on the X-ray transmission image. The grade rank is, for example, "S size," "M size," "L size," etc., and is set based on the weight range. For example, if the weight range is 10 to 20 g, it is set to M size, etc.

[0016] Before inspection, item A is carried into the X-ray inspection device 10 by the carry-in conveyor 60. After inspection, item A is carried out of the X-ray inspection device 10 by the carry-out conveyor 61. Item A determined to be a defective item by the X-ray inspection device 10 is sorted out of the production line by the sorting device 30 located downstream of the carry-out conveyor 61. Item A determined to be a non-defective item by the X-ray inspection device 10 passes through the sorting device 30 as is.

[0017] The device main body 2 houses a control unit 9 and the like. The support legs 3 support the device main body 2. The shielding box 4 is provided on the device main body 2. The shielding box 4 prevents leakage of X-rays (electromagnetic waves) to the outside. Inside the shielding box 4 is provided an inspection area R where inspection of item A is carried out using X-rays. The shielding box 4 is formed with an inlet 4a and an outlet 4b. An item A before inspection is carried into the inspection area R from the inlet 4a of the carry-in conveyor 60. An item A after inspection is carried out from the inspection area R to the outlet 4b of the carry-out conveyor 61. An X-ray shielding curtain (not shown) is provided at each of the inlet 4a and the outlet 4b to prevent leakage of X-rays.

[0018] The transport unit 5 is disposed so as to penetrate the center of the shielding box 4. The transport unit 5 transports the article A along the transport direction D from the entrance 4a through the inspection area R to the exit 4b. The transport unit 5 is, for example, a belt conveyor stretched between the entrance 4a and the exit 4b. Note that the transport unit 5, which is a belt conveyor, may protrude outward beyond the entrance 4a and the exit 4b.

[0019] As shown in Figures 2 and 3, the X-ray irradiator 6 is disposed within the shield box 4. The X-ray irradiator 6 irradiates X-rays onto the article A being transported by the transport unit 5. The X-ray irradiator 6 includes, for example, an X-ray tube that emits X-rays and a diaphragm that spreads the X-rays emitted from the X-ray tube in a fan shape in a plane perpendicular to the transport direction D. The X-rays irradiated from the X-ray irradiator 6 include X-rays in various energy bands ranging from low energy (long wavelength) to high energy (short wavelength). Note that the terms "low" and "high" in the low energy band and high energy band described above indicate relatively "low" and "high" among the multiple energy bands irradiated from the X-ray irradiator 6, and do not indicate a specific range.

[0020] The X-ray detection unit 7 is disposed within the shielding box 4. The X-ray detection unit 7 detects X-rays in each of a plurality of energy bands that have passed through the article A. In this embodiment, the X-ray detection unit 7 is configured to detect X-rays in a low-energy band and X-rays in a high-energy band. That is, the X-ray detection unit 7 has a first line sensor 11 and a second line sensor 12. The first line sensor 11 and the second line sensor 12 are each composed of X-ray detection elements arranged one-dimensionally along a horizontal direction perpendicular to the conveying direction D. The first line sensor 11 detects X-rays in the low-energy band that have passed through the article A and the conveying belt of the conveying unit 5. The second line sensor 12 detects X-rays in the high-energy band that have passed through the article A, the conveying belt of the conveying unit 5, and the first line sensor 11.

[0021] As shown in Fig. 2, the display operation unit 8 is provided on the device main body 2. The display operation unit 8 displays various information and accepts input of various conditions. The display operation unit 8 is, for example, a liquid crystal display, and displays an operation screen as a touch panel. In this case, the operator can input various conditions via the display operation unit 8.

[0022] The control unit 9 is disposed within the apparatus main body 2. The control unit 9 controls the operation of each part of the X-ray inspection apparatus 10. The control unit 9 includes processors such as a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit), memories such as a ROM (Read Only Memory) and a RAM (Random Access Memory), and storage such as an SSD (Solid State Drive). A program for controlling the X-ray inspection apparatus 10 is recorded in the ROM.

[0023] As shown in FIG. 4, the control unit 9 includes a communication unit 13, an image generation unit 14, a storage unit 15, a control unit 16, and an output unit 17.

[0024] The communication unit 13 communicates with the machine learning device 50. The communication unit 13 receives data of the trained model transmitted from the machine learning device 50. The trained model received by the communication unit 13 is stored in the storage unit 15. The communication unit 13 transmits the image data stored in the storage unit 15 to the machine learning device 50.

[0025] The image generation unit 14 generates an X-ray transmission image (X-ray image). The image generation unit 14 receives the X-ray detection results from the X-ray detection unit 7. The image generation unit 14 receives the low-energy band X-ray detection results from the first line sensor 11 (see FIG. 2) of the X-ray detection unit 7 and the high-energy band X-ray detection results from the second line sensor 12 (see FIG. 2) of the X-ray detection unit 7. The image generation unit 14 generates an X-ray transmission image based on these detection results. The image generation unit 14 outputs image data relating to the X-ray transmission image to the control unit 16. The image generation unit 14 stores the image data relating to the X-ray transmission image in the memory unit 15.

[0026] The memory unit 15 stores various information (data). The memory unit 15 stores the trained model output from the communication unit 20. The memory unit 15 stores X-ray transmission images (image data) generated by the image generation unit 14. The memory unit 15 also stores virtual X-ray images. The virtual X-ray images are images of virtual objects that are not the subject of inspection. The virtual object is, for example, the same type of object as object A, and the virtual X-ray images are, for example, X-ray transmission images generated in advance by passing X-rays through actual object A. However, the virtual object may be an object of a type other than object A. Furthermore, the virtual X-ray images may include virtual objects generated using computer graphics technology without passing X-rays through an actual object. The virtual X-ray images may be X-ray transmission images generated by the X-ray inspection apparatus 10, or may be acquired via a network or a storage medium, etc.

[0027] The control unit 16 inputs the X-ray image (image data) generated by the image generation unit 14 into the trained model and estimates the weight or grade rank of the item A shown in the X-ray image. The trained model includes, for example, a neural network NW. The trained model may include a convolutional neural network (CNN) or a Transformer. Furthermore, the trained model may include a neural network with multiple layers (e.g., eight or more layers). In other words, the trained model may be generated by deep learning.

[0028] As shown in FIG. 5, the neural network NW is composed of, for example, a first layer which is an input layer, a second layer, a third layer, and a fourth layer which are intermediate layers (hidden layers), and a fifth layer which is an output layer. The first layer receives an input value x=(x0, x1, x2, ... x) with p parameters as elements. p ) is output to the second layer as is. The second, third, and fourth layers each convert the total input into an output using an activation function and pass the output to the next layer. The fifth layer also converts the total input into an output using an activation function, and this output is the output value y=(y0, y1, ..., y q )

[0029] In this embodiment, the neural network NW is designed to input the pixel values ​​of each pixel in the inspection image and cluster and output the weight or grade rank of item A. The input layer of the neural network NW is provided with neurons equal to the number of pixels in the inspection image. The output layer of the neural network NW is provided with neurons for outputting the cluster results (weight range, grade rank) of the weight or grade rank of item A. Based on the output values ​​of the neurons in the output layer, information indicating whether the weight or grade rank of item A is good or bad can be determined.

[0030] The control unit 16 generates a sorting signal based on the estimated weight or grade of item A. The control unit 16 generates the sorting signal based on the output value of the neural network NW. The control unit 16 generates the sorting signal based on the cluster (group) indicated by the output value of the neural network NW for the weight or grade of item A. If the weight or grade of item A does not belong to a predetermined cluster, the control unit 16 generates a sorting signal including information indicating that the item is a defective item (defective item flag). The control unit 16 outputs the generated sorting signal to the output unit 17. As a result, when performing weight inspection, the control unit 16 generates a sorting signal so that if the estimated weight of item A is within a certain target allowable weight range, the item is classified as a non-defective item, and if it is not, the item is classified as a defective item. Furthermore, when performing rank sorting, the control unit 16 generates a sorting signal so that if the estimated weight of item A is within a weight range corresponding to each predetermined rank, the item is classified as a product of that rank, and if it is not within either range, the item is classified as a defective item.

[0031] However, if the control unit 16 cannot generate a sorting signal by the time the weight or grade rank of the item A, whose weight or grade rank has been estimated, reaches the sorting device 30, it exceptionally causes the output unit to output a signal instructing the sorting device 30 to discharge the item A. The control unit 16 obtains the time until the item A reaches the sorting device 30, for example, based on the conveying speed of the conveying unit 5 and the distance between the X-ray inspection device 10 and the sorting device 30. If the control unit 16 cannot generate a sorting signal by the time the weight or grade rank of the item A, whose weight or grade rank has been estimated, reaches the sorting device 30, it generates a sorting signal including information indicating that the item is an uninspected item (an uninspected flag).

[0032] When the conveyance unit 5 starts conveying the item A, the control unit 16 inputs a virtual X-ray image into the trained model and performs virtual processing to estimate the weight or grade rank of the virtual item. The start of conveyance of the item A in the conveyance unit 5 can occur when the X-ray inspection device 10 is turned on, or when inspection is resumed after being stopped in the X-ray inspection device 10, etc. When the conveyance unit 5 starts conveying the item A, the control unit 16 acquires a virtual X-ray image from the memory unit 15, inputs the virtual X-ray image into the trained model, and estimates the weight or grade rank of the virtual item. In the virtual processing, the control unit 16 does not generate a sorting signal regardless of whether the estimation result of the weight or grade rank of the virtual item is good or bad.

[0033] The output unit 17 outputs the sorting signal to the sorting device 30. The output unit 17 outputs the sorting signal output from the control unit 16 to the sorting device 30.

[0034] 1, the sorting device 30 is provided downstream of the X-ray inspection device 10. The sorting device 30 is provided on the carry-out conveyor 61. The sorting device 30 sorts the articles A based on a sorting signal output from the X-ray inspection device 10. The sorting device 30 has, for example, an arm (not shown).

[0035] The sorting device 30 may include a first discharge section (not shown) and a second discharge section (not shown). The sorting device 30 discharges the item A to the first discharge section or the second discharge section by operating an arm. The first discharge section accommodates, for example, defective items. The second discharge section accommodates, for example, uninspected items. If the sorting signal output from the X-ray inspection device 10 contains information indicating a defective item (defective item flag), the sorting device 30 operates the arm to discharge the item A to the first discharge section. As a result, the defective item A is discharged to the first discharge section. If the sorting signal output from the X-ray inspection device 10 contains information indicating an uninspected item (uninspected item flag), the sorting device 30 operates the arm to discharge the item A to the second discharge section. As a result, the uninspected item A is discharged to the second discharge section.

[0036] When sorting by rank based on grade rank, the sorting device 30 may be configured to sort the item A in a sorting direction according to each rank that is the inspection result of the item A. In this case, the control unit 16 outputs a sorting signal according to each rank that is the inspection result of the item A to the sorting device 30.

[0037] (When generating a trained model) Next, the configuration when generating a trained model will be described. The weighing device 40 weighs item A. The weighing device 40 weighs item A to be inspected by the X-ray inspection device 10. The weighing device 40 may be located upstream or downstream of the X-ray inspection device 10 on the transport path of item A. The weighing device 40 outputs measurement data related to the weight of item A to the machine learning device 50. At this time, training data that is a set of an X-ray transmission image including item A generated when item A passes through the X-ray inspection device 10 and data related to the weight or grade rank of item A generated when item A passes through the weighing device 40 is output to the machine learning device 50. Note that the weighing device 40 only needs to be temporarily installed when generating the trained model, and may be removed after the trained model is generated.

[0038] The machine learning device 50 is a device that generates a trained model through machine learning. The machine learning device 50 is configured with a CPU, a GPU, a ROM, a RAM, etc. The X-ray inspection device 10 and the machine learning device 50 are communicably connected via a wired or wireless network such as the Internet or a telephone network, and can send and receive information.

[0039] The machine learning device 50 includes a communication unit 51, a learning model generation unit 52, and a storage unit 53.

[0040] The communication unit 51 communicates with the X-ray inspection device 10 and the weighing device 40. The communication unit 51 receives image data transmitted from the X-ray inspection device 10 and outputs it to the learning model generation unit 52 and the storage unit 53. The communication unit 51 transmits data of the trained model generated in the learning model generation unit 52 to the X-ray inspection device 10. The communication unit 51 receives measurement data transmitted from the weighing device 40 and outputs it to the storage unit 53.

[0041] The learning model generation unit 52 acquires training data to be used for machine learning and performs machine learning using the acquired training data to generate a trained model. The training data includes images and other data. The images are X-ray transmission images taken by the X-ray inspection device 10. The other data are measurement data of item A weighed by the weighing device 40. The measurement data is data related to the weight or grade rank of item A shown in the X-ray transmission image. The other data also includes item information indicating the item in the image. The learning model generation unit 52 generates a trained model using a pair of the X-ray transmission image and the weight or grade rank of item A shown in the X-ray transmission image as training data.

[0042] The learning model generation unit 52 performs machine learning using each pixel value of the image as an input value to the neural network, and processing information corresponding to the image as an output value of the neural network, to generate (configure) a neural network NW. When pixel values ​​are used as input values, they are used as input values ​​for neurons associated with each pixel (the position of the pixel on the image). The above machine learning itself can be performed in the same way as conventional machine learning algorithms. The learning model generation unit 52 stores the generated trained model in the storage unit 53.

[0043] (Effects of the present invention) As described above, the X-ray inspection system 1 according to this embodiment stores a trained model generated using, as training data, a pair of an X-ray transmission image generated by the image generation unit 14 of the X-ray inspection device 10 and the weight or grade rank of the item A shown in the X-ray transmission image. The X-ray inspection device 10 inputs the X-ray transmission image of the item A being transported by the transport unit 5 into the trained model, and estimates the weight or grade rank of the item A. In this way, the X-ray inspection device 10 uses the trained model generated using the training data, thereby eliminating the need to adjust parameters to accommodate differences in sensors, etc., the characteristics of the item, etc. Therefore, the X-ray inspection device 10 does not require an operator to adjust parameters. Therefore, the X-ray inspection device 10 can simplify its setup.

[0044] In the X-ray inspection system 1 according to this embodiment, the trained model is generated using training data that is a pair of an X-ray transmission image and a weight or grade rank based on a weighing signal output from a weighing device 40 that detects the weight of the item A. In this configuration, the trained model is generated using training data that includes the accurate weight or grade rank of the item A actually weighed by the weighing device 40, thereby improving the estimation accuracy of the trained model. As a result, the X-ray inspection device 10 can improve the inspection accuracy.

[0045] In the X-ray inspection system 1 according to this embodiment, the storage unit 15 of the X-ray inspection device 10 stores a virtual X-ray image showing a virtual item that is not the target of inspection. When the transport unit 5 starts transporting item A, the control unit 16 inputs the virtual X-ray image into the trained model and performs virtual processing to estimate the weight or grade rank of the virtual item. When the transport unit 5 starts transporting item A, such as when the X-ray inspection device 10 is powered on, the control unit 16 has not yet performed the estimation process, so the initial estimation process may take some time. For example, the control unit 16 (specifically, the GPU, etc. included in the control unit 16) may be in a state where its processing power is temporarily reduced due to power saving or other reasons due to its specifications. This may result in the control unit 16 being unable to generate a sorting signal by the time the item A, whose weight or grade rank has been estimated, arrives at the sorting device 30.

[0046] Therefore, in the X-ray inspection apparatus 10, the control unit 16 inputs a virtual X-ray image into the trained model when the conveyance unit 5 starts conveying an item, and performs virtual processing to estimate the weight or grade rank of the virtual item. This allows the control unit 16 to improve its processing capacity in order to inspect the virtual X-ray image, so that the control unit 16 has a sufficiently increased processing capacity when inspecting the subsequent X-ray inspection image including the item A that is actually to be inspected. In other words, when inspecting the X-ray inspection image including the item A that is actually to be inspected, the control unit 16's processing capacity is insufficient, and a sorting signal cannot be generated before the item A reaches the sorting device 30, thereby reducing the occurrence of a situation in which the item A cannot be properly sorted. In this way, the X-ray inspection apparatus 10 performs virtual processing once before actually inspecting the item A, so that the item A that is actually to be inspected can be properly processed.

[0047] In the X-ray inspection system 1 according to this embodiment, if the control unit 16 is unable to generate a sorting signal before the weight or grade of the item A, whose weight or class rank has been estimated, reaches the sorting device 30, the control unit 16 outputs a signal from the output unit to the sorting device 30 instructing the item A to be discharged. In this configuration, if the sorting signal cannot be generated in time, the item A is forcibly discharged (discharged to the second discharge section). This makes it possible to prevent defective items from being sent to a subsequent process. In other words, it is possible to prevent an item A for which an incorrect inspection result was obtained from being transported as a non-defective item.

[0048] Although the embodiments of the present invention have been described above, the present invention is not necessarily limited to the above-described embodiments, and various modifications are possible without departing from the spirit of the present invention.

[0049] In the above embodiment, an example has been described in which the machine learning device 50 generates a trained model through machine learning, and the X-ray inspection device 10 performs processing using the trained model. However, the trained model may be generated by the X-ray inspection device 10. In other words, the X-ray inspection device may have the functionality of a machine learning device.

[0050] In the above embodiment, an example has been described in which the X-ray detection unit 7 in the X-ray inspection apparatus 10 has the first line sensor 11 and the second line sensor 12. However, the X-ray detection unit 7 may have only one line sensor.

[0051] In the above embodiment, the sorting device 30 is described as having a first discharge unit and a second discharge unit. However, the sorting device 30 may be configured to have only one discharge unit. In this case, defective products and uninspected products are configured to be discharged to this one discharge unit.

[0052] The X-ray detection unit 7 may also be capable of detecting X-rays using a photon counting method. The X-ray detection unit 7 may be a direct conversion type detection unit or an indirect conversion type detection unit. The X-ray detection unit 7 includes, for example, a sensor (multi-energy sensor) that detects X-rays in each of multiple energy bands that pass through the item A. The sensors are, for example, arranged in a direction (width direction) perpendicular to at least the conveying direction and the up-down direction of the conveying unit 5. The elements may be arranged not only in the width direction but also in the conveying direction. In other words, the X-ray detection unit 7 may include a line sensor or a group of sensors arranged two-dimensionally. The sensor is, for example, a photon detection type sensor such as a CdTe semiconductor detector. [Explanation of symbols]

[0053] 5...conveying unit, 10...X-ray inspection device (inspection device), 14...image generating unit, 15...storage unit, 16...control unit, 17...output unit, 30...sorting device, 40...weighing device.

Claims

1. a conveying unit that conveys the article; an image generating unit that captures an image of the object being transported by the transport unit and generates an X-ray image; a storage unit that stores a trained model generated using training data that is a pair of the X-ray image generated by the image generation unit and the weight or class rank of the item shown in the X-ray image; a control unit that inputs the X-ray image of the item being transported in the transport unit into the trained model, estimates the weight or the grade rank of the item, and generates a sorting signal based on the estimated weight or the grade rank of the item; an output unit that outputs the sorting signal to a sorting device that is provided downstream in the conveying direction of the item in the conveying unit.

2. 2. The inspection device according to claim 1, wherein the trained model is generated using as the training data a pair of the X-ray image and the weight or grade rank based on a weighing signal output from a weighing device that detects the weight of the item.

3. the storage unit stores a virtual X-ray image showing a virtual object that is not an inspection target; The inspection device of claim 1 or 2, wherein the control unit inputs the virtual X-ray image into the trained model when the transport unit starts transporting the item, and performs virtual processing to estimate the weight or the grade rank of the virtual item.

4. The inspection device described in claim 1 or 2, wherein the control unit outputs a signal instructing the sorting device to eject the item from the output unit if the sorting signal cannot be generated before the item, whose weight or grade rank has been estimated, reaches the sorting device.

Images