Learning device, learning method, and learning program
An intermediate image generating unit 81 generates a plurality of intermediate images from a good product image representing a good product and a faulty product image representing a faulty product. An intermediate image display unit 82 arranges the plurality of intermediate images between the good product image and the faulty product image and displays the same on a display device. A boundary acceptance unit 83 accepts, as a boundary between the good product image and the faulty product image, a designation by a user of a boundary between the intermediate images. A teaching image identification unit 84 identifies the images of the good product image and the faulty product image on the basis of the designated boundary.
1. A learning device comprising:
at least one memory configured to store instructions; and
at least one processor configured to execute the instructions, to perform:
generating a plurality of intermediate images from a good product image representing a good product and a faulty product image representing a faulty product;
arranging the plurality of intermediate images between the good product image and the faulty product image and displays the same on a display device;
accepting, as a boundary between the good product image and the faulty product image, a designation by a user of a boundary between the intermediate images; and
identifying the images of the good product image and the faulty product image on the basis of the designated boundary.
2. The learning device according to claim 1 , the instructions further comprising:
generating intermediate images in which the degrees of good product images or faulty product images are designated by parameters.
3. The learning device according to claim 1 , the instructions further comprising:
accepting the designation by the user of the position between the displayed intermediate images; and
identifying images on both sides of the designated position as a good product image and a faulty product image.
4. The learning device according to claim 1 , the instructions further comprising:
accepting the designation by the user of an intermediate image parameter; and
identifying the good product image and the faulty product image with the intermediate image generated by the designated parameter as a boundary.
5. The learning device according to claim 1 , the instructions further comprising:
generating a plurality of intermediate images from a good product image, a first faulty product image, and a second faulty product image;
arranging the first faulty product image and the second faulty product image in two axial directions different from each other, with the good product image as a point of origin, and displays the same on the display device;
accepting a designation by the user of a boundary between the good product image and the first faulty product image and of a boundary between the good product image and the second faulty product image; and
identifying the images of the good product image and the faulty product images on the basis of the designated boundaries.
6. The learning device according to claim 1 , the instructions further comprising:
relearning a model by using the good product image and faulty product image identified from the intermediate images.
7. A learning method comprising:
generating a plurality of intermediate images from a good product image representing a good product and a faulty product image representing a faulty product;
arranging the plurality of intermediate images between the good product image and the faulty product image and displaying the same on a display device;
accepting, as a boundary between the good product image and the faulty product image, a designation by a user of a boundary between the intermediate images; and
identifying the images of the good product image and the faulty product image on the basis of the designated boundary.
8. The learning method according to claim 7 , comprising
generating intermediate images in which the degrees of good product images or faulty product images are designated by parameters.
9. A non-transitory computer readable information recording medium configured to store a learning program: when executed by a processor, that performs a method for:
generating a plurality of intermediate images from a good product image representing a good product and a faulty product image representing a faulty product;
an intermediate image display process of arranging the plurality of intermediate images between the good product image and the faulty product image and displaying the same on a display device;
accepting, as a boundary between the good product image and the faulty product image, a designation by a user of a boundary between the intermediate images; and
a teaching image identification process of identifying the images of the good product image and the faulty product image on the basis of the designated boundary.
10. The non-transitory computer readable information recording medium according to claim 9 , further comprising: generating intermediate images in which the degrees of good product images or faulty product images are designated by parameters.