Improvement of the inspection training tool and the verification of the accuracy of the machine learning discriminant model using the results

Abstract

Recently, almost all the quality inspection work in the manufacturing industry has become automated. However, there are many products for which inspection cannot be automated. For example, all parts are slightly different because the tip of a dental treatment rotating tool (Diamond bar) is attached to diamond particles. In addition, judgments of the inspection are different by the operator. Our previous studies challenged the development of the inspection support tool. However, the tool's precision did not improve even if the various parameter adjustments were performed. In this study, we focused on the dependability of the sample data. Moreover, the distinction models are structured and inspect the performance. As a result, the new model using corrected samples was a higher performance than the using all samples.