Substrate defect inspection apparatus, substrate defect inspection method, and storage medium

摘要

The present invention provides a technology capable of appropriately classifying a defect in a substrate based on an image of an inspection target obtained by photographing the substrate. Binarized data including a defective region is obtained from an image obtained by photographing the substrate by a deep learning and two-class classification unit (3). A rectangular color image including a defective region is created from the binarized data, and a deep learning system is used to estimate the type of a first defect based on this image. Furthermore, attributes such as the shape of the defective region and the position on a wafer (W) are obtained based on the binarized data, and a rule-based system is used to estimate the type of a second defect based on these pieces of information. The type of defect is comprehensively determined based on the estimation result by the deep learning system and the estimation result by the rule-based system.