Image processing device, method for image processing, and image processing program

Abstract

To provide an image processing device, a method for image processing, and an image processing program that allow pseudo images necessary for improving accuracy of a trained model to be easily generated.SOLUTION: The image processing device includes: an acquisition unit for acquiring incorrect images in which prediction results of a trained model are incorrect; a classification unit for classifying the incorrect images acquired by the acquisition unit into one of a plurality of groups on the basis of features of the incorrect images; a display unit for displaying the incorrect images included in the group; a reception unit for receiving an instruction for generation of a pseudo image; and an image generation unit for generating the pseudo image on the basis of the instruction received by the reception unit.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present disclosure relates to an image processing device, an image processing method, and an image processing program. [Background technology]

[0002] Patent Document 1 discloses a learning data generation device that includes a defective part data storage unit that stores defective part data, which is image data of defective part images; a good product data storage unit that stores good product data, which is image data of good product images; a learning data generation unit that generates learning data, which is image data of learning images in which good product images are combined with defective part images, based on the defective part data, good product data, and generation parameters; a learning data storage unit that stores the learning data; and a generation parameter setting unit that sets the generation parameters. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2020-27424 Summary of the Invention [Problem to be solved by the invention]

[0004] One method for inspecting an item involves inputting an image containing the item into a trained model generated by machine learning, and inspecting the item according to the prediction results output from the trained model. In order to improve prediction accuracy, it is preferable to generate a trained model using a large amount of training data. However, in an actual inspection process, defective products are not often generated, making it difficult to obtain many defective images (actual images) of defective products. Therefore, it takes time to obtain many defective images, and therefore it takes time to generate a trained model. Therefore, as described above, pseudo images of defective images are generated, and machine learning of the training model is performed using the pseudo images as training data.

[0005] In order to improve the accuracy of a trained model, it is necessary to perform machine learning on the trained model using images that include features that the trained model is not good at (features that are difficult to predict as defects). Therefore, in order to instruct the generation of pseudo images that include features that the trained model is not good at, an operator must visually confirm images for which the trained model's prediction results are incorrect (mispredicted) (hereinafter referred to as "incorrect images") and characterize the incorrect images. However, even if an operator visually confirms incorrect images, it is difficult to characterize the incorrect images. Furthermore, even if an incorrect image is characterized, it is not easy to instruct the generation of pseudo images that have features that the trained model is not good at based on the features.

[0006] The present disclosure aims to provide an image processing device, an image processing method, and an image processing program that can easily generate pseudo-images necessary to improve the accuracy of a trained model. [Means for solving the problem]

[0007] An image processing device according to one aspect of the present disclosure includes an acquisition unit that acquires incorrect images for which the prediction results of a trained model are incorrect, a classification unit that classifies the incorrect images into one of a plurality of groups based on the characteristics of the incorrect images acquired by the acquisition unit, a display unit that displays the incorrect images included in the group, a reception unit that receives instructions to generate pseudo images, and an image generation unit that generates pseudo images based on the instructions received by the reception unit. [Effects of the Invention]

[0008] According to the present disclosure, it is possible to easily generate pseudo images necessary to improve the accuracy of a trained model. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a diagram illustrating an example of the hardware configuration of a computer used in a prediction system according to an embodiment. [Figure 2]FIG. 2 is a diagram illustrating an example of a functional configuration of the prediction system. [Figure 3] FIG. 3 is a diagram showing an example of the result of clustering the incorrect images. [Figure 4] FIG. 4 is a diagram showing an example of a screen displayed on the output device. [Figure 5] FIG. 5 is a diagram showing an example of a screen displayed on the output device. [Figure 6] FIG. 6 is a diagram showing an example of a screen for instructing the generation of a pseudo image. [Figure 7] FIG. 7 is a diagram illustrating an example of a neural network. [Figure 8] FIG. 8 is a flowchart showing an example of a process flow for generating a pseudo image. [Figure 9] FIG. 9 is a flowchart showing an example of a learning method and generation of a trained model as a processing flow. [Figure 10] FIG. 10 is a flowchart showing an example of a prediction method as a processing flow. DETAILED DESCRIPTION OF THE INVENTION

[0010] [Description of the embodiments of the present disclosure] First, the contents of the embodiments of the present disclosure will be listed and described. At least some of the embodiments described below may be combined in any combination.

[0011] (1) An image processing device according to one aspect of the present disclosure includes an acquisition unit that acquires incorrect images for which the prediction results of a trained model are incorrect; a classification unit that classifies the incorrect images into one of a plurality of groups based on the characteristics of the incorrect images acquired by the acquisition unit; a display unit that displays the incorrect images included in the group; a reception unit that receives instructions to generate pseudo images; and an image generation unit that generates pseudo images based on the instructions received by the reception unit.

[0012] In an image processing device according to one aspect of the present disclosure, incorrect images are classified into one of a plurality of groups based on the characteristics of the incorrect images, and the incorrect images included in the group are displayed on a display unit. This allows an operator to check the incorrect images classified into groups by their characteristics on the display unit, thereby easily grasping the characteristics of the incorrect images for each group. Therefore, the operator can easily issue instructions for generating pseudo images based on the group. Therefore, the image processing device can easily generate pseudo images necessary for improving the accuracy of the trained model.

[0013] (2) The image processing device of (1) above may further include a verbalization unit that verbalizes the characteristics of the incorrect images included in each of the multiple groups classified by the classification unit, and the display unit may display text of the characteristics verbalized by the verbalization unit. In this configuration, the characteristics of the incorrect images included in the group can be confirmed in text. Therefore, the operator can more easily recognize the characteristics of the incorrect images.

[0014] (3) In the image processing device of (1) or (2) above, the display unit may display an incorrect image that best represents the characteristics of the incorrect images in the group among the multiple incorrect images included in the group. The incorrect image best represents the characteristics of the incorrect images in the group. Therefore, by checking the incorrect image, the worker can recognize the characteristics of the incorrect images in the group at a glance.

[0015] (4) In any one of the image processing devices described in (1) to (3), the reception unit may receive a selection of a group of pseudo images to be generated by the image generation unit and a designation of the size of the pseudo images, and the image generation unit may generate the pseudo images based on the group and size received by the reception unit. With this configuration, the pseudo images desired by the operator can be easily generated.

[0016] (5) An image processing method according to one aspect of the present disclosure includes an acquisition step of acquiring an incorrect image for which the prediction result of a trained model is incorrect; a step of classifying the incorrect image into one of a plurality of groups based on the characteristics of the incorrect image acquired in the acquisition step; a display step of displaying the incorrect image included in the group; a reception step of receiving an instruction to generate a pseudo image; and an image generation step of generating a pseudo image based on the instruction received in the reception step.

[0017] In an image processing method according to one aspect of the present disclosure, incorrect images are classified into one of a plurality of groups based on the characteristics of the incorrect images, and the incorrect images included in the group are displayed on a display unit. This allows the operator to check the incorrect images classified into groups by their characteristics on the display unit, thereby easily grasping the characteristics of the incorrect images for each group. Therefore, the operator can easily issue instructions for generating pseudo images based on the group. Therefore, the image processing method makes it possible to easily generate pseudo images necessary for improving the accuracy of a trained model.

[0018] (6) An image processing program according to one aspect of the present disclosure causes a computer to execute an acquisition step of acquiring an incorrect image for which the prediction result of a trained model is incorrect; a step of classifying the incorrect image into one of a plurality of groups based on the characteristics of the incorrect image acquired in the acquisition step; a display step of displaying the incorrect image included in the group; a reception step of receiving an instruction to generate a pseudo image; and an image generation step of generating a pseudo image based on the instruction received in the reception step.

[0019] An image processing program according to one aspect of the present disclosure classifies incorrect images into one of a plurality of groups based on the characteristics of the incorrect images, and displays the incorrect images included in the group on a display unit. This allows an operator to check the incorrect images classified into groups by their characteristics on the display unit, thereby easily grasping the characteristics of the incorrect images for each group. Therefore, the operator can easily issue instructions for generating pseudo images based on the group. Therefore, the image processing program can easily generate pseudo images necessary for improving the accuracy of a trained model.

[0020] [Details of the embodiments of the present disclosure] Specific examples of embodiments of the present disclosure will be described below with reference to the drawings. The present disclosure is not limited to these examples, but is defined by the claims, and is intended to include all modifications within the meaning and scope of the claims. In the description of the drawings, the same elements are given the same reference numerals, and duplicate explanations will be omitted.

[0021] The prediction system 1 according to this embodiment is, for example, a computer system that predicts the presence or absence of defects in the inspection of an item. The inspection of the item is, for example, an appearance inspection of the item. In this embodiment, the result predicted by the prediction system 1 is referred to as the "prediction result."

[0022] The prediction system 1 uses machine learning to predict the presence or absence of defects. Machine learning is a technique for autonomously finding laws or rules by iteratively (repeatedly) learning based on given information. The prediction system 1 performs machine learning using a machine learning model. For example, the prediction system 1 may perform machine learning using a gradient boosting decision tree (GBDT), machine learning using multiple regression analysis, or machine learning using a convolutional neural network (CNN) configured to include a convolutional layer and a pooling layer. A convolutional neural network is a type of deep learning that uses a neural network with a multilayer structure.

[0023] The prediction system 1 trains a machine learning model by repeating learning, and acquires this machine learning model as a trained model. This corresponds to the learning phase. In the learning phase, the prediction system 1 functions as a learning device (image processing device). The trained model is a machine learning model that is predicted to be optimal for predicting the presence or absence of defects. The prediction system 1 processes input data using the trained model, and outputs a prediction result of the presence or absence of defects, which corresponds to the operation phase (prediction phase). In the operation phase, the prediction system 1 functions as a prediction device.

[0024] Trained models are portable between computer systems. Therefore, a trained model generated in one computer system can be used in another computer system. Of course, one computer system may execute both the generation and use of trained models. That is, the prediction system 1 may execute both the learning phase and the operation phase, or may not execute either the learning phase or the operation phase. In this embodiment, the prediction system 1 executes both the learning phase and the operation phase.

[0025] In the learning phase, the prediction system 1 uses training data. The training data is image data. The image data may include good product images, defective images, and pseudo images. Good product images are images that do not contain defective parts. Defective images are images that include defective parts. Pseudo images include pseudo good product images and pseudo defective images. The prediction system 1 generates a trained model by performing machine learning using the training data. In the operation phase, the prediction system 1 obtains prediction results by providing input data to the trained model. The input data is image data captured during an inspection process, etc.

[0026] [System Configuration] FIG. 1 is a diagram illustrating an example of a general hardware configuration of a computer 100 constituting the prediction system 1. For example, the computer 100 includes a processor 101, a main memory 102, an auxiliary memory 103, a communication control unit 104, an input device 105, and an output device (display) 106. The processor 101 executes an operating system and application programs. The main memory 102 is composed of, for example, ROM and RAM. The auxiliary memory 103 is composed of, for example, a hard disk or flash memory, and generally stores a larger amount of data than the main memory 102. The communication control unit 104 is composed of, for example, a network card or a wireless communication module. The input device 105 is composed of, for example, a keyboard, a mouse, a touch panel, etc. The output device 106 is composed of, for example, a monitor and a speaker.

[0027] Each functional element of the prediction system 1 is realized by a learning program 110 and a prediction program 120 pre-stored in the auxiliary storage unit 103. Specifically, each functional element is realized by loading the learning program 110 or the prediction program 120 onto the processor 101 or the main storage unit 102 and executing the learning program 110 or the prediction program 120. The processor 101 operates the communication control unit 104, the input device 105, or the output device 106 in accordance with the learning program 110 or the prediction program 120, and reads and writes data from and to the main storage unit 102 or the auxiliary storage unit 103. Data or databases required for processing are stored in the main storage unit 102 or the auxiliary storage unit 103.

[0028] The learning program 110 and the prediction program 120 may be provided by being fixedly recorded on a tangible recording medium such as a CD-ROM, a DVD-ROM, a semiconductor memory, etc. Alternatively, the learning program 110 and the prediction program 120 may be provided via a communication network as a data signal superimposed on a carrier wave.

[0029] The prediction system 1 may be configured with one computer 100 or multiple computers 100. When multiple computers 100 are used, these computers 100 are connected via a communication network such as the Internet or an intranet, thereby logically constructing a single prediction system 1.

[0030] Fig. 2 is a diagram illustrating an example of the functional configuration of the prediction system 1. As illustrated in Fig. 2, the prediction system 1 includes an acquisition unit 10, a clustering unit (classification unit) 11, a verbalization unit 12, a storage unit 13, a reception unit 14, an image generation unit 15, a learning unit 16, a prediction unit 17, and a database 20.

[0031] The acquisition unit 10 is a functional element that acquires image data. The acquisition unit 10 acquires image data from a database 20. The acquisition unit 10 can access the database 20 that stores the image data. The database 20 can be used to train a learning model. The database 20 may be, for example, a component of the prediction system 1, or may be built in a computer system separate from the prediction system 1. The prediction system 1 and the database 20 may be connected via a communication network, or both the prediction system 1 and the database 20 may be built in a single computer.

[0032] The method for preparing the image data to be stored in the database 20 is not limited. For example, the image data may be stored in the database 20 by an operator's input work, or may be automatically collected and stored in the database 20 by the prediction system 1 or another computer system. The database 20 stores, as image data, good product images, bad product images, pseudo images, and incorrect product images. An incorrect product image is an image for which the prediction result of the trained model is incorrect. The incorrect product images may include good product images and bad product images. The acquisition unit 10 acquires the incorrect product images from the database 20 as image data. The acquisition unit 10 outputs the acquired image data to the clustering unit 11.

[0033] The clustering unit 11 is a functional element that clusters incorrect images. The clustering unit 11 clusters the incorrect images into one of a plurality of groups based on the features of the incorrect images acquired by the acquisition unit 10. The number of groups can be set to, for example, 3 to 5. The number of groups can be set as appropriate. The clustering unit 11 clusters the incorrect images based on the similarity of their features. The clustering unit 11 inputs the incorrect images into a trained model, and clusters the incorrect images into a predetermined group based on the output result of the trained model. The trained model is a machine learning model that clusters input images into set groups.

[0034] FIG. 3 is a diagram showing an example of the results of clustering incorrect images. In the example shown in FIG. 3, the incorrect images are clustered into five groups. In FIG. 3, group A is indicated by "●", group B is indicated by "○", group C is indicated by "▲", group D is indicated by "△", and group E is indicated by "■". As shown in FIG. 3, the incorrect images are clustered based on similar features.

[0035] As shown in FIG. 2, the verbalization unit 12 is a functional element that verbalizes the features of an incorrect image. For each of the multiple groups classified by the clustering unit 11, the verbalization unit 12 verbalizes the features of the incorrect images included in the group. The verbalization unit 12 inputs the incorrect images included in each group into a trained model to acquire the features of the incorrect images and verbalize the features. The trained model acquires (extracts) the features based on a predetermined prompt. The predetermined prompt may be set by input by an operator or may be set automatically. The predetermined prompt may be, for example, "color tone of the image," "texture of the image," "shape of a defective part (foreign substance)," etc. The trained model is a machine learning model that extracts features from an input image based on the prompt. A known machine learning model can be used as the trained model. The verbalization unit 12 stores the acquired features in the storage unit 13.

[0036] The memory unit 13 is a functional element that stores data, etc. The memory unit 13 stores the features verbalized by the verbalization unit 12 for each category. The categories correspond to the content of the prompt. That is, the categories can be, for example, "color of image," "texture of image," "shape of defective part (foreign substance)," etc. The memory unit 13 stores text indicating the features for each category. For example, the memory unit 13 stores text such as "black" or "blue" for "color of image," and stores text such as "round" or "triangle" for "shape of defective part." The features stored in the memory unit 13 can be confirmed by an operator on, for example, the output device 106 (monitor).

[0037] The output device 106 displays the incorrect images included in the group. Of the multiple incorrect images included in the group, the output device 106 displays the incorrect image that best represents the characteristics of the incorrect images of the group. FIG. 4 is a diagram showing an example of a screen displayed on the output device 106. As shown in FIG. 4, the output device 106 displays the clustering results and a representative image of each group (the incorrect image that best represents the characteristics).

[0038] When a representative image of a group is selected (designated), the output device 106 displays text describing the characteristics of the incorrect images included in the group. Fig. 5 is a diagram showing an example of a screen displayed on the output device 106. As shown in Fig. 5, when group A is selected, for example, text describing the characteristics of group A is displayed.

[0039] The reception unit 14 is a functional element that receives instructions (input) from an operator via the input device 105. The reception unit 14 receives instructions to generate pseudo images (pseudo-good images, pseudo-defective images). The reception unit 14 outputs instruction information related to the instructions to generate the pseudo images to the image generation unit 15.

[0040] As shown in FIG. 2, the image generation unit 15 is a functional element that generates a training image. The image generation unit 15 performs image processing on an image of the image data to generate a training image (pseudo image). The image generation unit 15 generates a pseudo-fault image based on instruction information output from the reception unit 14. FIG. 6 is a diagram showing an example of a screen for instructing the generation of a pseudo image. As shown in FIG. 6, an instruction screen for instructing the generation of a pseudo image is displayed on the output device 106. The instruction screen displays an item for selecting a group (A to E) and an item for specifying a size. The instruction screen also displays a representative image of the selected group. The worker operates the input device 105 to specify group A and size, and presses a button to start generation.

[0041] The image generation unit 15 generates a pseudo-fault image including a defective part based on an instruction received by the reception unit 14. The image generation unit 15 generates a pseudo-fault image based on the group and size received by the reception unit 14. The image generation unit 15 generates a pseudo-fault image based on features corresponding to the group stored in the storage unit 13. The image generation unit 15 inputs the features into a trained model to generate a pseudo-fault image. The trained model is a machine learning model (image generation AI) that generates an image based on text.

[0042] The learning unit 16 is a functional element that performs machine learning. The learning unit 16 performs machine learning of a neural network. The learning unit 16 performs machine learning of a learning model using images stored in the database 20. The learning unit 16 inputs learning data into the input layer of the neural network, and repeatedly (iteratively) performs a learning process on the neural network to generate a trained model. The learning unit 16 stores the trained model obtained by repeating the learning process a predetermined number of times in the memory unit 13.

[0043] The prediction unit 17 is a functional element that predicts the presence or absence of a defect using a trained model. The prediction unit 17 inputs input data to the trained model. In response to inputting the input data to the neural network NW (see FIG. 7) of the trained model, the prediction unit 17 acquires a prediction result including an output value output from the neural network NW.

[0044] FIG. 7 is a diagram showing an example of a neural network NW. As shown in FIG. 7, the neural network NW is composed of, for example, a first layer which is an input layer L1, a second layer, a third layer, and a fourth layer which are intermediate layers (hidden layers) L2, and a fifth layer which is an output layer L3. The first layer outputs an input value x=(x0, x1, x2, ... xp) having p+1 parameters as elements to the second layer as is. Each of the second, third, and fourth layers converts the total input into an output using an activation function and passes 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, ... , yp) of the neural network NW having q+1 parameters as elements. q )

[0045] In this embodiment, the neural network NW receives the pixel values ​​of each pixel of image data as input data and outputs information indicating a prediction result. The input layer L1 of the neural network NW is provided with neurons equal to the number of pixels in the image. The output layer L3 of the neural network NW is provided with neurons for outputting information related to the prediction result of the target object. The presence or absence of a defect can be predicted based on the output values ​​(likelihoods) of the neurons in the output layer.

[0046] [Operation of the prediction system] (Learning phase) The image generation method will be described with reference to Fig. 8. Fig. 8 is a flowchart showing an example of pseudo image generation as a processing flow S1. Processing flow S1 corresponds to a learning phase.

[0047] In step S11, the acquisition unit 10 acquires incorrect images from the database 20. In step S12, the clustering unit 11 clusters the incorrect images into one of a plurality of groups based on the features of the incorrect images acquired by the acquisition unit 10. In step S13, for each of the plurality of groups classified by the clustering unit 11, the features of the incorrect images included in the group are verbalized.

[0048] In step S14, the output device 106 displays the clustering results and text of the features. In step S15, the reception unit 14 receives an instruction to generate a pseudo image. In step S16, the image generation unit 15 generates a pseudo image based on the group and size received by the reception unit 14.

[0049] The learning method and the method for generating a trained model will be described with reference to Fig. 9. Fig. 9 is a flowchart showing an example of the learning method and the generation of a trained model as a processing flow S2. The processing flow S2 corresponds to the learning phase.

[0050] In step S21, the acquisition unit 10 acquires images from the database 20. In step S22, the learning unit 16 performs machine learning using the training images. In step S23, the learning unit 16 determines whether to terminate learning. If the termination condition for machine learning is met, the learning unit 16 terminates learning, and if the termination condition is not met, the learning unit 16 continues machine learning. The termination condition is set arbitrarily. For example, the termination condition may be set based on the error from the correct answer, or may be set based on the number of training data to be processed, i.e., the number of times learning is performed.

[0051] If learning is to continue (step S23: NO), the learning unit 16 acquires the next training image and performs the processes from step S22 onwards for that training image. If learning is to end (step S23: YES), the learning unit 16 acquires a trained model in step S24. Thus, in the learning phase, the prediction system 1 generates a trained model by performing machine learning using training data.

[0052] (Operational phase) The prediction method will be described with reference to Fig. 10. Fig. 10 is a flowchart showing an example of the prediction method as a processing flow S3. The processing flow S3 corresponds to the operation phase and is an example of the prediction method according to the present disclosure.

[0053] In step S31, the prediction unit 17 acquires input data. In step S32, the prediction unit 17 inputs the processed data into a trained model and outputs a prediction result obtained by the trained model. The method for outputting the prediction result by the prediction unit 17 is not particularly limited. For example, the prediction unit 17 may output the prediction result to the output device 106, store it in a predetermined database, or transmit it to another computer system.

[0054] [effect] As described above, the prediction system 1 according to this embodiment classifies incorrect images into one of multiple groups based on the characteristics of the incorrect images, and displays the incorrect images included in the group on the display unit. This allows the worker to check the incorrect images classified into groups by their characteristics on the display unit, making it easy to understand the characteristics of the incorrect images for each group. This allows the worker to easily issue instructions for generating pseudo images based on the group. Therefore, the prediction system 1 can easily generate pseudo images necessary to improve the accuracy of the trained model.

[0055] The prediction system 1 according to this embodiment includes a verbalization unit 12 that verbalizes the characteristics of the incorrect images included in each of the multiple groups classified by the clustering unit 11. The output device 106 displays text of the characteristics verbalized by the verbalization unit 12. In this configuration, the characteristics of the incorrect images included in the group can be confirmed in text. Therefore, the worker can more easily recognize the characteristics of the incorrect images.

[0056] In the prediction system 1 according to this embodiment, the output device 106 displays an incorrect image that best represents the characteristics of the incorrect images in the group among the multiple incorrect images in the group. The incorrect image best represents the characteristics of the incorrect images in the group. Therefore, by checking the incorrect image, the worker can recognize the characteristics of the incorrect images in the group at a glance.

[0057] In the prediction system 1 according to this embodiment, the input device 105 accepts the selection of a group of pseudo images to be generated by the image generation unit 15 and the specification of the size of the pseudo images, and the image generation unit 15 generates the pseudo images based on the group and size accepted by the input device 105. With this configuration, the operator can easily generate the pseudo images he or she desires.

[0058] 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.

[0059] In the above embodiment, the prediction system 1 is described as including the output device 106 as a display unit and the input device 105 as a reception unit. However, the display unit and the reception unit may be a touch panel display having two functions.

[0060] In the above embodiment, the neural network NW is described as being composed of a first layer, which is an input layer L1, a second layer, a third layer, and a fourth layer, which are intermediate layers (hidden layers) L2, and a fifth layer, which is an output layer L3. However, the neural network may include one or more intermediate layers. [Explanation of symbols]

[0061] 1. Prediction system 10…Acquisition part 11...Clustering Department 12…Language department 13...Storage section 14. Reception 15...Image generation unit 16...Study Department 17...Prediction Department 20...Database 100...Computer 101...Processor 102...Main memory section 103…Auxiliary storage unit 104...Communication control unit 105...Input device 106...Output device 110...Study Program 120...Prediction program A...Group B...Group C...Group D...Group E...Group L1...input layer L2…middle layer L3: Output layer NW...neural network

Claims

1. an acquisition unit that acquires incorrect images for which the prediction results of the trained model are incorrect; a classification unit that classifies the incorrect image into one of a plurality of groups based on the characteristics of the incorrect image acquired by the acquisition unit; a display unit that displays the incorrect images included in the group; a reception unit that receives an instruction to generate a pseudo image; an image generation unit that generates the pseudo image based on the instruction received by the reception unit.

2. a verbalization unit that verbalizes, for each of the plurality of groups classified by the classification unit, characteristics of the incorrect images included in the group; The image processing device according to claim 1 , wherein the display unit displays text of the features verbalized by the verbalization unit.

3. The image processing device according to claim 1 , wherein the display unit displays, from among the plurality of incorrect images included in the group, the incorrect image that best represents the characteristic of the incorrect images of the group.

4. the receiving unit receives a selection of the group of pseudo images to be generated by the image generating unit and a designation of a size of the pseudo images; The image processing device according to claim 1 , wherein the image generating unit generates the pseudo image based on the group and the size received by the receiving unit.

5. an acquisition step of acquiring incorrect images for which the prediction results of the trained model are incorrect; classifying the incorrect image into one of a plurality of groups based on the characteristics of the incorrect image acquired in the acquiring step; a display step of displaying the incorrect images included in the group; a receiving step of receiving an instruction to generate a pseudo image; an image generating step of generating the pseudo image based on the instruction received in the receiving step.

6. an acquisition step of acquiring incorrect images for which the prediction results of the trained model are incorrect; classifying the incorrect image into one of a plurality of groups based on the characteristics of the incorrect image acquired in the acquiring step; a display step of displaying the incorrect images included in the group; a receiving step of receiving an instruction to generate a pseudo image; an image generating step of generating the pseudo image based on the instruction received in the receiving step.

Images