Data Enhancement for Convolution Neural Network-Based Defect Inspection

摘要

A system and method for providing enhanced input data to a convolutional neural network (CNN) is disclosed. The wafer image is received at the processor. The wafer image is divided into a plurality of reference images each associated with a die in a wafer image. A test image is received. A plurality of difference images are generated by obtaining the difference between the test image and the reference image. The reference image and the difference image are aggregated into augmented input data for CNN and provided to CNN.