Computer-implemented method for detection and classification of anomalies in imaging datasets of wafers and system using same

A computer-implemented method (28, 28 ') for detecting and classifying anomalies (15) in an imaging data set (66) of a wafer comprising a plurality of semiconductor structures is disclosed. The method includes determining current detections of a plurality of anomalies (15) in the imaging data set (66) and obtaining an unsupervised or semi-supervised grouping of the current detections of the plurality of anomalies (15). At least one of the groups is selected for presentation and annotation to the user via a user interface (236) according to at least one decision criterion. The anomaly classification algorithm is retrained in accordance with anomalies (15) of the annotations. A system (234) for controlling wafer quality and a system (234 ') for controlling wafer production are also disclosed. 本发明揭示一种用于检测以及分类包括多个半导体结构的晶片的成像数据集(66)中异常(15)的计算机实施方法(28、28')。该方法包含：确定该成像数据集(66)中多个异常(15)的当前检测，以及获得该多个异常(15)的当前检测的无监督或半监督成群。根据至少一个决策标准，选择该成群中至少一个集群，以经由一使用者界面(236)给使用者呈现且注记。该异常分类算法根据注记的异常(15)进行重新训练。本发明另揭示一种用于控制晶片质量的系统(234)、以