Method and apparatus for creating learning materials for constructing a cutting edge wear determination model for cutting tools.

Abstract

A method for generating learning data for machine learning, which uses multiple photographic images (PIs) of the cutting edge (CE) of a cutting tool to construct a loss determination model of the cutting edge (CE), is characterized by: photographing the cutting edge (CE) to obtain multiple photographic images (PIs); performing dimensional compression on the multiple photographic images (PIs) to extract two-dimensional or three-dimensional feature quantities (FAs); depicting the feature quantities (FAs) extracted by dimensional compression in a feature space (FS2, FS3); identifying blank regions (ERs) of feature quantity (FA) data in the feature space (FS2, FS3) based on the density of the depiction in the feature space (FS2, FS3); and generating a generated image from the photographic images (PIs) for extracting feature quantities (FAs), so as to supplement the specified blank regions (ERs) with feature quantity (FA) data.

Images