COMPUTER PROGRAM, GENERATION DEVICE, AND GENERATION METHOD
To appropriately generate data for inspection.SOLUTION: A generation device detects, from each of P (P is an integer of 2 or more) pieces of first training image data each representing a photographic image of a first type physical object including K (K is an integer of 1 or more) objects and having...
Saved in:
Main Author | |
---|---|
Format | Patent |
Language | English Japanese |
Published |
16.02.2023
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | To appropriately generate data for inspection.SOLUTION: A generation device detects, from each of P (P is an integer of 2 or more) pieces of first training image data each representing a photographic image of a first type physical object including K (K is an integer of 1 or more) objects and having no abnormality in its appearance, K object areas corresponding to the K objects. The generation device uses a plurality of images of a first attention object area of the K object areas to execute, on each of the K object areas, processing of training a first attention machine learning model being a machine learning model associated with the first attention object area. The generation device generates first correspondence data indicating K types of correspondence between first object area information and first model identification information and corresponding respectively to the K object areas. The generation device stores K trained machine learning models and the first correspondence data in a storage.SELECTED DRAWING: Figure 1
【課題】 検査用のデータを適切に生成する【解決手段】K個(Kは、1以上の整数)のオブジェクトを含み外観の異常の無い第1種対象物の撮影画像をそれぞれ表すP個(Pは、2以上の整数)の第1訓練画像データのそれぞれから、K個のオブジェクトに対応するK個のオブジェクト領域を検出する。K個のオブジェクト領域のうちの第1注目オブジェクト領域の複数の画像を使用して、第1注目オブジェクト領域に対応付けられる機械学習モデルである第1注目機械学習モデルを訓練する処理を、K個のオブジェクト領域のそれぞれに対して実行する。第1オブジェクト領域情報と第1モデル識別情報との対応関係であって、K個のオブジェクト領域にそれぞれ対応するK個の対応関係を示す第1対応関係データを生成する。K個の訓練済の機械学習モデルと、第1対応関係データとを、記憶装置に格納する。【選択図】 図1 |
---|---|
Bibliography: | Application Number: JP20210129596 |