Inspection device, inspection method, machine learning device, and machine learning method
The purpose of the present invention is to improve accuracy in inspection of an appearance of a mold. The inspection device includes at least one processor for performing an inspection step of inspecting an appearance of a mold using a learned model constructed by machine learning. Input into the learned model includes an inspection image obtained by imaging the appearance of the mold. Output from the learned model is information indicating an inspection result of the appearance of the mold.
1. An inspection device, comprising:
at least one processor for performing an inspection step of inspecting a mold using a learned model constructed by machine learning,
input into the learned model being an inspection image obtained by imaging the mold and a reference image obtained by imaging a normal mold, and,
output from the learned model being information indicating at least one of a presence or an absence of a defect of the mold, a defect region of the mold, or both the presence or the absence of a defect of the mold and the defect region of the mold, wherein:
in the inspection step, the at least one processor
reads the inspection image and the reference image from a memory,
calculates the information from the inspection image and the reference
image with use of the learned model, and,
writes the information in the memory.
2. The inspection device according to claim 1 , wherein: the image is an image of a product surface of the mold.
3. The inspection device according to claim 1 , wherein:
the at least one processor further performs a selection step of selecting a learned model according to a type of a pattern used to produce the mold; and, in the inspection step, the at least one processor inspects the mold using the learned model selected in the selection step.
4. The inspection device according to claim 1 , wherein: the input into the learned model further includes one or more or all of sand information indicating foundry sand constituting the mold, molding information indicating a molding status of the mold, conveyance information indicating a conveyance status of the mold, and environment information.
5. The inspection device according to claim 1 , wherein: the output from the learned model indicates at least presence or absence of a defect in the mold.
6. The inspection device according to claim 5 , wherein: if the output from the learned model indicates that the mold has a defect, the at least one processor further performs a warning step of outputting warning information based on information indicating the inspection result.
7. An inspection method, comprising:
an inspection step of inspecting mold using a learned model constructed by machine learning, the inspection step being performed by at least one processor,
input into the learned model being an inspection image obtained by imaging the mold and a reference image obtained by imaging a normal mold, and,
output from the learned model being information indicating at least one of a presence or an absence of a defect of the mold, a defect region of the mold, or both the presence or the absence of a defect of the mold and the defect region of the mold, wherein:
in the inspection step, the at least one processor
reads the inspection image and the reference image from a memory,
calculates the information from the inspection image and the reference
image with use of the learned model, and,
writes the information in the memory.
8. A machine learning device for constructing a learned model for inspecting a mold, comprising:
at least one processor for performing a construction step of constructing, by supervised learning with a dataset-for-learning, a learned model for inspecting a mold,
input into the learned model being an inspection image obtained by imaging the mold and a reference image obtained by imaging a normal mold, the processor calculating information from the inspection image and the reference image with use of the learned model, and,
output from the constructed learned model being the information indicating at least one of a presence or an absence of a defect of the mold, a defect region of the mold, or both the presence or the absence of a defect of the mold and the defect region of the mold, wherein:
in the construction step, the at least one processor
reads the dataset-for-learning from a memory,
sets parameters defining the learned model with use of the dataset-for-learning, and
writes the learned model in the memory.
9. A machine learning method for constructing a learned model for inspecting a mold, comprising:
a construction step of constructing, by supervised learning with a dataset-for-learning, a learned model for inspecting a mold, the construction step being performed by at least one processor,
input into the learned model being an inspection image obtained by imaging the mold and a reference image obtained by imaging a normal mold the processor calculating information from the inspection image and the reference image with use of the learned model, and,
output from the constructed learned model being the information indicating at least one of a presence or an absence of a defect of the mold, a defect region of the mold, or both the presence or the absence of a defect of the mold and the defect region of the mold, wherein:
in the construction step, the at least one processor
reads the dataset-for-learning from a memory,
sets parameters defining the learned model with use of the dataset-for-learning, and
writes the learned model in the memory.