Abnormality inspection device and abnormality inspection method

1. An abnormality inspection device comprising processing circuitry

to acquire a determination target image in which a target object to be inspected is captured;

to acquire a result of machine learning of forward domain conversion of an image or inverse domain conversion of an image performed using a normal image in which the target object in a normal state is captured as training data;

to obtain a domain-converted image by sequentially performing forward domain conversion and inverse domain conversion on the determination target image, using the result of the machine learning;

to determine whether or not an abnormality is occurring in the target object captured in the determination target image by comparing the determination target image and the domain-converted image;

to obtain a forward-domain-converted image by performing forward domain conversion on the determination target image using a result of machine learning performed using the normal image as training data; and

to perform analysis on the abnormality using the forward-domain-converted image when the abnormality is determined to be occurring in the target object captured in the determination target image.

2. The abnormality inspection device according to claim 1 , wherein the processing circuitry further performs:

to acquiring the normal image; and

to perform the machine learning using the normal image.

3. The abnormality inspection device according to claim 1 ,

wherein the forward domain conversion or the inverse domain conversion is performed using a neural network.

4. The abnormality inspection device according to claim 1 ,

wherein the determination unit generates a difference image whose pixels indicate absolute values of differences between a plurality of pixels included in the determination target image and a plurality of pixels included in the domain-converted image, which positionally correspond to each other, and determines an area formed by a set of pixels having the absolute values greater than or equal to a threshold value among the pixels of the difference image as an area where an abnormality is occurring in the target object captured in the determination target image.

5. The abnormality inspection device according to claim 1 , wherein the processing circuitry further performs:

to output, to a display device, information regarding a determination result of whether or not the abnormality is occurring in the target object captured in the determination target image.

6. An abnormality inspection method comprising:

acquiring a determination target image in which a target object to be inspected is captured;

acquiring a result of machine learning of forward domain conversion of an image or inverse domain conversion of an image performed using a normal image in which the target object in a normal state is captured as training data;

obtaining a domain-converted image by sequentially performing forward domain conversion and inverse domain conversion on the determination target image, using the result of the machine learning;

determining whether or not an abnormality is occurring in the target object captured in the determination target image by comparing the determination target image and the domain-converted image;

obtaining a forward-domain-converted image by performing forward domain conversion on the determination target image using a result of machine learning performed using the normal image as training data; and

performing analysis on the abnormality using the forward-domain-converted image when the abnormality is determined to be occurring in the target object captured in the determination target image.