機械学習装置および該方法ならびに膜密着性評価装置および該方法

Japanese Patent JP7393321

B2

To provide a machine learning device and machine learning method that enable automatization, as well as a film adhesion evaluation device and film adhesion evaluation method for evaluating the adhesion of a film and equipped with the device and method.SOLUTION: A machine learning device pertaining to the present invention comprises a training data acquisition unit 1 for acquiring training data, and a machine learning execution unit 32 for training a machine learning model on the basis of the training data. The training data includes an indentation image in which the indentation of Rockwell indentation test is imaged and a label, as teaching data, that represents the degree of adhesion of a film associated with the indentation image, the label being represented by a numerical value including decimal parts, the machine learning model outputting a label for the image that is inputted. A film adhesion evaluation device D is equipped with the machine learning device, and includes an image acquisition unit 2 for acquiring a target image obtained by imaging the film to be evaluated, where an indentation of Rockwell indentation test is formed, and evaluation processing units 33, 34 for inputting the target image to the machine learning model having been trained by machine learning and obtaining a label as the evaluation result.SELECTED DRAWING: Figure 1

Kosuke Umetani

JP2020206969A

December 06, 2023

December 14, 2020

Click for automatic bibliography generation  

KABUSHIKI KAISHA KOBE SEIKO SHO

G01N19/04; G01N3/42; G06T7/00

JP5006354U
JP2020060879A
JP2019141927A
JP2011102712A

WO2019219455A1

Masataka Kotani
Satoshi Sakurai

Previous Patent: siphon drainage system

Next Patent: Temperature measurement device, temperature measurement method and atmospheric measurement system