Positive/Negative Decision Via Outlier Detection Towards Automatic Performance Evaluation for Defect Detector

Abstract

In the field of semiconductor defect inspection, it has been possible to detect defects with high accuracy thanks to the object detection model (defect detector) composed of the deep learning model. The performance of the deep learning model depends highly on training data; therefore, during the operational phase at the customer site, we need to frequently evaluate the model's performance to deal with shifts of appearance for defects. However, frequently executing general evaluation methods is difficult at the customer site; hence, we need a method to automatically evaluate performance. In this study, for the purpose of automatically evaluating the performance of the defect detector, we propose the Positive/Negative Decision via Outlier Detection (PNDOD). PNDOD decides on positive/negative for detection results based on comparing features corresponding to the detected result with statistics computed from training data. By using this method, we can calculate the estimated precision from the ratio of the estimated number of positive detections to the number of total detections, and we can evaluate the model performance automatically based on this estimated precision. In experiments using SiC wafer images, we confirmed that PNDOD can decide on positive/negative with high accuracy, and we can precisely evaluate the model's performance.