AI Support System

Abstract

To provide a user-friendly AI support system which reduces the number of inspection targets to be visually inspected by workers.SOLUTION: An AI support system comprises a photographed image obtaining unit 11, a quality determination unit 2, a display control unit 13, and a storage unit 12. The quality determination unit 2 includes: a first determination unit 22 which excludes a photographed image G in which an inspection target is determined to be normal; and a second determination unit 23 which performs clustering of the inspection target based on the photographed image G in which the inspection target is determined to be defective by the first determination unit 22. The display control unit 13 displays on a worker terminal 6 the photographed image G clustered by the second determination unit 23 and defect classification of the inspection target.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to an AI support system for determining whether an image is good or bad on a production line in a factory or the like. [Background technology]

[0002] Conventionally, an AI support system is known that acquires images of inspection objects taken by a camera installed on a production line via a network and determines the quality of the inspection objects using an AI model (see, for example, Patent Document 1).

[0003] The AI ​​support system described in Patent Document 1 displays the judgment result of normal on the worker's terminal when it judges that the inspection object is normal, and displays the judgment result of normal and an image of the inspection object on the worker's terminal when it judges that the inspection object is defective. Also, if the inspection object is defective, the judgment result and the image are displayed on the worker's terminal, and the worker tags the cause of the defect and re-learns the AI ​​model. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] International Publication No. 2022 / 124316 Summary of the Invention [Problem to be solved by the invention]

[0005] The AI ​​support system described in Patent Document 1 only displays the judgment result on the worker's terminal when the inspection object is normal, which reduces the consumption of communication resources. However, when the inspection object is, for example, a nonwoven fabric that flows sequentially from a roll to the production line, the number of inspection objects becomes enormous. As a result, the number of inspection objects that workers must visually inspect is large, and there is room for improvement.

[0006] Furthermore, the AI ​​support system described in Patent Document 1 displays the judgment result and the captured image on the worker's terminal when the inspection object is found to be defective, and the worker tags the cause of the defect and re-learns the AI ​​model. However, when there are a wide variety of causes for defects in the inspection objects, the worker needs specialized knowledge to tag the causes of the defects, so there is room for improvement in terms of usability.

[0007] Therefore, there is a need for an easy-to-use AI support system that reduces the number of items that workers need to visually inspect. [Means for solving the problem]

[0008] The AI ​​support system according to the present invention is characterized by comprising an image acquisition unit that acquires images of an inspection object; a pass / fail judgment unit that judges whether the inspection object is pass / fail based on the captured images; a display control unit that controls the display screen of an operator terminal; and a memory unit that stores the captured images for which the inspection object is judged to be defective by the pass / fail judgment unit, wherein the pass / fail judgment unit has a first judgment unit that excludes the captured images for which the inspection object is judged to be normal, and a second judgment unit that clusters the inspection object based on the captured images for which the inspection object is judged to be defective by the first judgment unit, and the display control unit displays the captured images clustered by the second judgment unit and the defect classification of the inspection object on the operator terminal.

[0009] The pass / fail judgment unit in this configuration has a first judgment unit that excludes captured images for which the inspection object is judged to be normal, and is provided with a memory unit that stores captured images for which the inspection object is judged to be defective.Therefore, even if there are a huge number of inspection objects that are normal, the storage capacity can be reduced by not storing captured images that are judged to be normal.

[0010] Furthermore, the pass / fail judgment unit in this configuration has a second judgment unit that performs clustering of inspection objects based on captured images in which the first judgment unit has judged the inspection objects to be defective. In a work site where there is a huge number of inspection objects that are judged to be normal, the second judgment unit narrows down the captured images in which the first judgment unit has judged the inspection objects to be defective and performs clustering of the inspection objects, thereby significantly increasing processing speed.

[0011] Furthermore, the display control unit in this configuration displays the captured images clustered by the second determination unit and the defect classification of the inspection objects on the worker terminal, allowing the worker to visually inspect the captured images with a significantly reduced number of inspection objects and correct the defect classification. As a result, the worker can determine and correct the defect classification of the inspection objects by looking only at the significantly reduced number of captured images, reducing the number of inspection objects that the worker must visually inspect, resulting in a user-friendly AI support system.

[0012] Another characteristic feature is that the memory unit stores the defect classification for the captured image in association with the captured image, and when the input classification input from the operator terminal differs from the defect classification, the pass / fail judgment unit re-learns based on the input classification.

[0013] In this configuration, the storage unit stores the defect classification for each captured image in association with the captured image, so that, for example, once a certain number of captured images have been accumulated, the captured images can be read out sequentially in a batch process, and an operator can correct the defect classification.Then, the pass / fail judgment unit re-learns based on the input classification entered by the operator, thereby improving the judgment accuracy of the AI ​​support system.

[0014] Another characteristic feature is that when the first judgment unit judges that the inspection object is normal, the display control unit does not display anything on the worker terminal, and when the second judgment unit judges that the inspection object is normal, the display control unit displays the probability of normality and the captured image on the worker terminal.

[0015] In this configuration, inspection objects determined to be normal by the first determination unit are excluded and nothing is displayed on the worker terminal, so when there are a huge number of inspection objects that are normal, the worker does not need to check anything, which greatly improves work efficiency. Also, only when the second determination unit determines that an inspection object is normal are the probability of normal and the captured image displayed on the worker terminal, so even if the worker checks this together with the captured image determined to be defective, the workload does not increase.

[0016] Another characteristic feature is that the memory unit stores the defect classification for the captured image in association with the captured image, and when the first judgment unit judges that the inspection object is defective, the display control unit displays thumbnails of the captured image and the attributes of the inspection object on the operator terminal, and displays a list of judgment probabilities of the defect classification in parallel with the thumbnail display.

[0017] In this configuration, the storage unit stores the defect classification for the captured image in association with the captured image, so that the worker can input the defect classification when he has time by batch processing. Also, the captured image and the attributes of the inspection object are displayed as thumbnails on the worker terminal, and a list of the judgment probabilities of the defect classification is displayed in parallel with the thumbnail display, so that even a skilled worker with no specialized knowledge can easily check the accuracy of the defect classification, thereby improving the work efficiency of the defect classification input work.

[0018] Another characteristic configuration is that the memory unit stores the defect classification for the captured image in association with the captured image, the pass / fail judgment unit calculates feature values ​​of the captured image for the inspection object judged to be defective by the first judgment unit, and the display control unit displays a collection of the feature values ​​for each defect classification in a two-dimensional space.

[0019] By displaying a collection of features for each defect classification in two-dimensional space as in this configuration, defect classifications with close feature distances can be easily understood. As a result, it becomes possible to determine whether the captured image used for re-learning is suitable, for example, if a defect A is re-learned but is located near defect B in two-dimensional space, it can be excluded from the re-learning target. [Brief explanation of the drawings]

[0020] [Figure 1] This is an overview diagram of the AI ​​support system. [Figure 2] FIG. 1 is a block diagram illustrating the configuration of an AI support system. [Figure 3] This is a control flow diagram of the AI ​​support system. [Figure 4A] This is an explanatory diagram of the re-learning of the AI ​​support system. [Figure 4B] This is a histogram before and after relearning. [Figure 5] This is an example of defect classification for inspection targets. [Figure 6] This is an example of the management screen of the AI ​​support system. [Figure 7] 10 is an example of a display screen of a worker terminal. [Figure 8] 10 is an example of a display screen of a worker terminal. [Figure 9] 10 is an example of a display screen of a worker terminal. DETAILED DESCRIPTION OF THE INVENTION

[0021] An embodiment of an AI support system according to the present invention will be described below with reference to the drawings. In this embodiment, an AI support system X that inspects nonwoven fabric products will be described as an example of an AI support system. However, the present invention is not limited to the following embodiment, and various modifications are possible within the scope of the gist of the present invention.

[0022] Fig. 1 is an overall explanatory diagram showing an image of the AI ​​support system X, and Fig. 2 is a block diagram showing a schematic configuration of the AI ​​support system X. As shown in Fig. 2, the AI ​​support system X is configured with the functional units of an image acquisition unit 11, a quality determination unit 2, a memory unit 12, and a display control unit 13. Each of these functional units is constructed of hardware or software, or both, with a processor such as a CPU as its core component, in order to perform processing related to determining defects such as scratches and dust on the surface of an inspection target (product) on a factory production line.

[0023] The image acquisition unit 11 acquires captured images G of nonwoven fabric products (an example of an inspection target, hereinafter referred to as "products") flowing through a factory production line. Nonwoven fabric products are made from raw fabric rolls fed from a roll, cut with a slit blade, and then shipped as merchandise. As shown in FIG. 1 , in this embodiment, to check the condition of these products, a camera 5 installed at a predetermined installation location above the product conveyance line sequentially captures images of the continuously flowing nonwoven fabric products. Note that the installation configuration of the camera 5 shown in FIG. 1 is merely an example, and the camera 5 may be installed in an installation configuration different from that shown in FIG. 1 . That is, the number of cameras 5 may be two or more, or may be one. Furthermore, the installation location may be a structure dedicated to installing the camera 5, or may be an existing structure already installed in the factory for another purpose. As shown in FIG. 2 , the image acquisition unit 11 receives multiple captured images G captured by the camera 5 via a communication line such as the Internet and transmits them to the quality determination unit 2. That is, the image acquisition unit 11 is a communication interface.

[0024] The quality determination unit 2 determines the quality of the product based on the captured image G acquired by the captured image acquisition unit 11. The quality determination unit 2 includes a feature calculation unit 21, a first determination unit 22, a second determination unit 23, a deviation map generation unit 24, and a re-learning unit 25. The quality determination unit 2 in this embodiment uses, for example, a neural network to map each feature in a feature space with a plot corresponding to each feature so that the higher the similarity between the features is, the shorter the distance is, and the lower the similarity is, the longer the distance is, and calculates the length of this distance as the deviation. In order to use such a neural network to appropriately calculate the degree of discrepancy between two feature quantities, the pass / fail judgment unit 2 is configured with a trained model that has been deep-learned using known metric learning or the like so that when two feature quantities with high similarity are input, the distance between the two plots corresponding to these two feature quantities in the feature quantity space becomes short, and when two feature quantities with low similarity are input, the distance between the two plots corresponding to these two feature quantities in the feature quantity space becomes long.

[0025] The feature calculation unit 21 calculates a feature (distance vector) of the input captured image G relative to a collection of normal captured images G, using a trained model that has undergone deep learning of multiple captured images G in which the product is in a normal or defective state. This feature may be a Euclidean distance or a Mahalanobis distance.

[0026] The first determination unit 22 excludes captured images G determined to be normal. Specifically, the first determination unit 22 determines that a captured image G is normal if the abnormality level based on the feature calculated by the feature calculation unit 21 is equal to or less than a predetermined threshold (e.g., 0.01). The first determination unit 22 does not transmit the captured image G to the second determination unit 23. As shown in FIG. 1, the first determination unit 22 outputs the feature calculated by the feature calculation unit 21 as an abnormality level based on a distance vector, ranging from 0 to 1, sets a threshold for captured images G with a 100% probability of being abnormal, and excludes captured images G with a value equal to or less than the threshold. This threshold may vary depending on product attributes (product number, etc.) and is pre-set as a boundary at which the probability of abnormality is 100% for a trained model that has been deep-learned for each product attribute. In the example shown in FIG. 1, a threshold of 0.4 results in the inclusion of defect data, so the threshold is set to 0.01. This setting may be set by an administrator visually inspecting the histogram, or may be automatically set to a threshold that excludes defect data.

[0027] The second determination unit 23 performs clustering of products based on captured images G determined by the first determination unit 22 to be defective. Specifically, the second determination unit 23 determines the probability of similarity with respect to a plurality of preset defect classifications based on the feature amounts (distance vectors) calculated by the feature amount calculation unit 21. As shown in FIG. 1 , the second determination unit 23 determines, for a captured image G determined to be defective, that the defect is defect A at a probability of 60%, defect B at a probability of 25%, and defect C at a probability of 15%, in ascending order of the feature amounts (distance vectors) for defects A to C, which are the defect classifications. In this embodiment, there are multiple defect classifications, and stains, lint, dust, tears, gouges, wrinkles, dirt, insects, etc. are set as defect classifications (see also FIG. 5 ). Note that the second determination unit 23 determines that a normal captured image G that was not excluded by the first determination unit 22 has a higher probability of being normal than any of the defect classifications.

[0028] The deviation map generation unit 24 generates a feature space (a two-dimensional space that is a collection of points obtained by plotting distance vectors on XY coordinates) for the feature amounts of the captured image G calculated by the feature calculation unit 21. Specifically, the deviation map generation unit 24 maps the feature amounts of the normal products excluded by the first determination unit 22 and the feature amounts of the products clustered by the second determination unit 23, for each normal and multiple defective classifications. As shown in FIG. 9 , the deviation map generation unit 24 generates a map in which the normal feature amounts and the feature amounts of defects A to D are plotted in two-dimensional space so that the distance between them is increased according to the respective deviations. This makes it possible to determine whether the captured image G used for relearning is suitable, for example, by excluding the captured image G from the relearning target when it is re-learned as defect A but is located near defect B in the two-dimensional space.

[0029] The re-learning unit 25 learns based on the input classification (including a defective classification and a normal classification) input from the operator terminal 6, which will be described later. Even if the defective classification determined by the second judgment unit 23 (including a case where the first judgment unit 22 judged the defect as defective but the second judgment unit 23 judged the defect as normal, the same applies below) is incorrect, re-learning can be performed based on the correct defective classification. Specifically, as shown in FIG. 4A , the second judgment unit 23 classifies defect A as defective because the probability that it is defect A is 60%, but if the operator visually checks the captured image G and finds that it is defect B, the defect A is corrected to the input defect B and the trained model is re-learned. As a result, the probability that it is defect B becomes 80% as the judgment result after re-learning. The re-learning unit 25 stores, for example, the captured image G and the corrected defect classification for each product attribute (product number) in the memory unit 12, and reads out the captured image G and its defect classification from the memory unit 12 when a certain amount of data has been accumulated or at predetermined intervals, and performs deep learning using known metric learning or the like to reconstruct the learned model.

[0030] 2, the storage unit 12 is configured with a storage medium such as an HDD or SSD that stores a trained model, a plurality of captured images G, and their attributes. In the present embodiment, the storage unit 12 stores, for each captured image G, its attributes, such as inspection date and time, feature amount, abnormality level (normal / failure classification), defect classification, product number, and its specifications (width, length, number of divisions, etc.). The storage unit 12 does not need to store the plurality of captured images G and their attributes that have been determined to be normal by the first determination unit 22, but it is preferable to store only the normal feature amount.

[0031] That is, the storage unit 12 stores the defect classification for the captured image G in association with the captured image G. This allows, for example, when a large number of captured images G have accumulated, the captured images G to be sequentially read out by batch processing, and enables an operator to perform defect classification determination and correction work when he or she has time. Then, the re-learning unit 25 performs re-learning based on the input classification entered by the operator, thereby improving the determination accuracy of the AI ​​support system X.

[0032] The display control unit 13 controls the display screen of the worker terminal 6, and causes the worker terminal 6 to display the captured images G clustered by the second determination unit 23 and the product defect classification. The worker terminal 6 is configured as a mobile terminal such as a personal computer or tablet that can be operated by a worker who performs inspection work in the factory. The display control unit 13 transmits the multiple captured images G and their attributes stored in the storage unit 12, as well as data related to the feature space generated by the deviation map generation unit 24, to the worker terminal 6, and causes it to be displayed on the screen.

[0033] When the first determination unit 22 determines that the product is normal, the display control unit 13 displays nothing on the worker terminal 6, but when the second determination unit 23 determines that the product is normal, the display control unit 13 displays the probability of normality (including the probability of other defective classifications) and the captured image G on the worker terminal 6. In this way, since nothing is displayed on the worker terminal 6 excluding products determined to be normal by the first determination unit 22, when there is a huge number of normal products such as nonwoven fabrics, the worker does not need to check anything, which significantly improves work efficiency. Furthermore, since the probability of normality and the captured image G are displayed on the worker terminal 6 only when the second determination unit 23 determines that the product is normal, the worker's workload is not increased even if they check these together with the captured image G determined to be defective.

[0034] 7, when the first judgment unit 22 judges that the product is defective, the display control unit 13 causes the worker terminal 6 to display thumbnails of the captured image G and the product attributes as the judgment results of the second judgment unit 23, and also causes a list of judgment probabilities of defective classification (including the probability of judgment as normal) to be displayed in parallel to the thumbnail display. In this way, because the worker terminal 6 displays thumbnails of the captured image G and the product attributes, and also displays a list of judgment probabilities of defective classification in parallel to the thumbnail display, even if the worker is not an experienced worker with specialized knowledge, it is possible to easily confirm the accuracy of the defective classification, and the work efficiency of the work of correcting the defective classification can be improved.

[0035] As described above, the pass / fail judgment unit 2 in this embodiment has a first judgment unit 22 that excludes captured images G for which the product has been judged to be normal, and is provided with a memory unit 12 that stores captured images G for which the product has been judged to be defective.Therefore, even in the case where there are a huge number of normal products, such as nonwoven fabrics, the storage capacity can be reduced by not storing captured images G that have been judged to be normal.

[0036] In addition, the pass / fail judgment unit 2 has a second judgment unit 23 that performs product clustering based on the captured images G in which the first judgment unit 22 has judged the product to be defective.In a work site where there are a huge number of products that are judged to be normal, the second judgment unit 23 narrows down the captured images G that are defective to perform product clustering, thereby significantly increasing processing speed.

[0037] Furthermore, the display control unit 13 displays the captured images G clustered by the second determination unit 23 and the defective product classification on the worker terminal 6, so that the worker can visually check the captured images G in which the number of products has been significantly reduced, and determine and correct the defective product classification. As a result, the worker can determine and correct the defective product classification by looking only at the captured images G in which the number has been significantly reduced, resulting in an easy-to-use AI support system X.

[0038] Next, the control process of the AI ​​support system X will be explained mainly using Figure 3. The control process described below is composed of a program executed by a computer, an application that runs on this program, and a non-transitory storage medium that stores the program or application.

[0039] When products start to flow on the product conveying line, the camera 5 sequentially captures images of the nonwoven fabric products as they flow. The captured image acquisition unit 11 then sequentially acquires the captured images G captured by the camera 5 (#31, captured image acquisition process). Next, the captured images G are transmitted to the quality determination unit 2 (trained model) via an internet line or the like, and the feature calculation unit 21 calculates the feature corresponding to each captured image G (feature calculation process). At this time, the feature calculation unit 21 converts the feature into a value between 0 and 1 as an abnormality degree, and calculates the abnormality degree (#32, abnormality calculation process). A histogram showing the distribution of the abnormality degree is shown in the upper left of FIG. 1, and the threshold is set to 0.01, for example.

[0040] Next, the first determination unit 22 determines whether the degree of abnormality calculated by the feature amount calculation unit 21 is equal to or less than a predetermined threshold (for example, 0.01) (#33, first determination process). If the degree of abnormality is equal to or less than the predetermined threshold, the first determination unit 22 determines that the captured image G is normal, does not transmit the captured image G to the second determination unit 23, transmits only the feature amount, and does not send anything to the worker terminal 6 or notify that it is normal (#33 YES, #40, #41). In other words, since products determined to be normal by the first determination unit 22 are excluded and nothing is displayed on the worker terminal 6, when there are a huge number of normal products such as nonwoven fabrics, the worker does not need to check anything, which significantly improves work efficiency.

[0041] On the other hand, if the degree of abnormality exceeds a predetermined threshold, the first judgment unit 22 judges the captured image G to be defective and transmits the captured image G and its features to the second judgment unit 23, and the memory unit 12 stores the captured image G and its features (#34, memory processing).

[0042] Next, the second determination unit 23 performs clustering of the products based on the captured images G in which the first determination unit 22 determined that the products are defective (#35, second determination process). FIG. 5 shows an example of clustering by the second determination unit 23. In case 1, a large depression is found in the nonwoven fabric, so it is determined to be a gouge. In case 2, small debris is found, so it is determined to be debris. In case 3, multiple scattered stains are found, so it is determined to be stains. These determinations are made automatically by a trained model that has undergone deep learning using metric learning or the like, so there is no need for an operator to visually track the product or make a judgment by looking at the captured images G.

[0043] When the second determination unit 23 performs clustering, it calculates the probability of determining a defective classification based on the degree of deviation from the determined defective classification (for example, the distance from the average value of features belonging to the same defective classification) (#36, second determination process). The defective classification and determination probability associated with the captured image G are stored in the memory unit 12 (storage process). FIG. 7 shows an example of the determination probability. As shown in the right diagram of FIG. 7, it has been determined that the determination probability of defect A is 96%, the determination probability of defect B is 2%, the determination probability of defect C is 1%, and the determination probability of defect D is 1%.

[0044] Furthermore, the deviation map generation unit 24 generates a feature space (a two-dimensional space that is a collection of points where distance vectors are plotted on XY coordinates) for the feature amounts of the captured image G calculated by the feature calculation unit 21 (deviation map generation process). An example of the deviation map is shown in Fig. 9. As shown in Fig. 9, point cloud maps are created for normal, defect A, defect B, defect C, and defect D, respectively, and it can be determined whether the captured image G is suitable for relearning.

[0045] Next, based on the defect classification (including determination of normality) determined by the second determination unit 23 and the captured image G, the display control unit 13 generates display screen data for the operator terminal 6 and transmits it to the operator terminal 6 (#37, display process). The operator terminal 6 displays a list screen as shown in FIG. 6. This list displays the number of captured images G (defect count) stored as defect classifications for each product number of the lot number, and indicates whether "slit registration" and "label registration" have been completed as statuses. "Slit registration" refers to information such as the length of the original web and the width of the slit processing. "Label registration" refers to the determination and correction of defect classifications. In this embodiment, as shown in FIG. 7, when performing label registration, the display control unit 13 displays thumbnails of the captured image G and product attributes (e.g., defect classification such as "Defect A") on the operator terminal 6 (left side of FIG. 7), and also displays a list of defect classification determination probabilities (right side of FIG. 7) in parallel with the thumbnail display. Clicking on a captured image G in this thumbnail display displays the list of defect classification determination probabilities. As a result, since the list of judgment probabilities of the defect classification is displayed in parallel with the thumbnail display, even an unskilled worker with specialized knowledge can easily check the accuracy of the defect classification, thereby improving the work efficiency of the defect classification correction work.

[0046] Next, the quality judgment unit 2 judges whether the worker has completed the defect classification and correction work for each product number of the lot number based on the status information of the "label registration" described above (re-learning judgment process). When the worker has completed the judgment work (#38), the re-learning unit 25 re-learns the trained model (#39, re-learning process). As shown in FIG. 4A, the re-learning unit 25 reads the captured image G and its defect classification from the memory unit 12 when a certain amount of data has accumulated or at predetermined intervals, and performs deep learning using well-known deep metric learning or the like to reconstruct the trained model. As a result, even if defect data (defects) and normal data are mixed overall as shown in the upper part of FIG. 4B, as a result of re-learning the defect data with normal data, the variation between the defect data (defects) and normal data is reduced, as shown in the lower part of FIG. 4B, and the judgment accuracy of the trained model is improved.

[0047] The display control unit 13 in this embodiment can display the performance of the current trained model, as shown in Fig. 8. As described above, because a threshold is set in the first determination unit 22, the probability that a defective image will be determined to be normal is 0%. As a result, if the degree of abnormality is equal to or less than the predetermined threshold, the first determination unit 22 determines that the captured image G is normal, and does not transmit anything to the operator terminal 6, nor does it notify the operator that the image is normal.

[0048] The "false positive detection rate (Precision)" is true positive / (true positive + false positive), the "recall rate (Recall)" is true positive / (true positive + false negative), and the "overall accuracy (Accuracy)" is (true positive + true negative) / total. In the example shown in FIG. 8, the "false positive detection rate (Precision)" is 75%, the "recall rate (Recall)" is 66%, and the "overall accuracy (Accuracy)" is 96%. In the example shown in FIG. 8, the "recall rate (Recall)" is a binary classification of normal / failure. Therefore, the normal detection rate is 100%, and the probability that the first determination unit 22 will determine a fault as normal is 0%. Setting a target value for the trained model (e.g., an overall accuracy of 95% or more) using these indices can also prevent overlearning by the re-learning unit 25.

[0049] 9, the display control unit 13 in this embodiment can map the feature amounts of normal products excluded by the first determination unit 22 and the feature amounts of products clustered by the second determination unit 23 for normal and multiple defect classifications. This makes it possible to determine whether the captured image G used for relearning is suitable, for example, if a defect A is relearned but is located near defect B in two-dimensional space, it can be excluded from the relearning target.

[0050] [Other embodiments] (a) In the above embodiment, the example of the inspection object is a nonwoven fabric, but the inspection object may be any product that requires inspection.

[0051] (b) Although the above-described re-learning unit 25 executes re-learning by batch processing, it may execute re-learning each time the input classification differs from the failure classification.

[0052] (c) In the above-described embodiment, when the second judgment unit 23 judges that the product is normal, the probability of normality and the captured image G are displayed on the worker terminal 6, but it is also possible that nothing needs to be displayed.

[0053] (d) In the above-described embodiment, when the first judgment unit 22 judges that a product is defective, the display control unit 13 causes the captured image G and the product attributes to be displayed as thumbnails on the worker terminal 6, and also causes a list of judgment probabilities of defective classification to be displayed in parallel with the thumbnail display. Alternatively, the thumbnail display may be omitted, or the list of judgment probabilities of defective classification may be displayed on a separate screen when a thumbnail-displayed captured image G is clicked.

[0054] (e) The browser used in the worker terminal 6 in the above-described embodiment is merely an example, and any suitable browser can be used as long as it does not deviate from the gist of the present invention. [Industrial Applicability]

[0055] The present invention can be used in an inspection system used in a production line in a factory or the like. [Explanation of symbols]

[0056] 2: Good / bad judgement section 6: Worker terminal 11: Image acquisition unit 12: Storage section 13: Display control section 22:First judgment part 23:Second judgment section G: Captured image X: AI Support System

Claims

1. an image acquisition unit that acquires an image of an inspection object; a quality determination unit that determines the quality of the inspection object based on the captured image; a display control unit that controls a display screen of the worker terminal; a storage unit that stores the captured image of the inspection object determined to be defective by the quality determination unit, the quality determination unit includes a first determination unit that excludes the captured images for which the inspection object is determined to be normal, and a second determination unit that performs clustering of the inspection object based on the captured images for which the inspection object is determined to be defective by the first determination unit, The display control unit displays the captured image clustered by the second judgment unit and the defect classification of the inspection object on the worker terminal, and when the first judgment unit judges that the inspection object is normal, does not display anything on the worker terminal, and when the second judgment unit judges that the inspection object is normal, displays the probability of normality and the captured image on the worker terminal. This is an AI support system.

2. An image acquisition unit that acquires an image of an inspection object; a quality determination unit that determines the quality of the inspection object based on the captured image; a display control unit that controls a display screen of the worker terminal; a storage unit that stores the captured image of the inspection object determined to be defective by the quality determination unit, the quality determination unit includes a first determination unit that excludes the captured images for which the inspection object is determined to be normal, and a second determination unit that performs clustering of the inspection object based on the captured images for which the inspection object is determined to be defective by the first determination unit, the display control unit causes the captured image clustered by the second determination unit and the defect classification of the inspection object to be displayed on the operator terminal, and when the inspection object is determined to be defective by the first determination unit, causes the captured image and the attributes of the inspection object to be displayed as thumbnails on the operator terminal, and also causes a list of determination probabilities of the defect classification to be displayed in parallel with the thumbnail display; The storage unit stores the defect classification for the captured image in association with the captured image.

3. An image acquisition unit that acquires an image of an inspection object; a quality determination unit that determines the quality of the inspection object based on the captured image; a display control unit that controls a display screen of the worker terminal; a storage unit that stores the captured image of the inspection object determined to be defective by the quality determination unit, The quality determination unit includes a first determination unit that excludes the captured images of the inspection object that have been determined to be normal, and a second determination unit that performs clustering of the inspection object based on the captured images of the inspection object that have been determined to be defective by the first determination unit, and calculates feature amounts of the captured images of the inspection object that have been determined to be defective by the first determination unit, the display control unit causes the captured image clustered by the second determination unit and the defect classification of the inspection object to be displayed on the operator terminal, and causes a collection of the feature amounts to be displayed for each defect classification in a two-dimensional space; The storage unit stores the defect classification for the captured image in association with the captured image.

4. the storage unit stores the defect classification for the captured image in association with the captured image, The AI ​​support system according to any one of claims 1 to 3, wherein the pass / fail judgment unit re-learns based on the input classification input from the operator terminal when the input classification differs from the failure classification.

Images