Training neural networks for defect detection in low resolution images

摘要

Methods and systems for training neural networks for detecting defects in low resolution images are provided.One system includes an inspection tool including a high - and low resolution imaging subsystem and one or more components including a high resolution neural network and a low resolution neural network.The computer subsystem of the system is configured to generate a defect image training set.At least one of the defect images is synthetically generated by a high resolution neural network using images generated by the high resolution imaging subsystem.The computer subsystem is also configured to train a low resolution neural network using a defective image training set as an input.In addition, the computer subsystem is configured to detect defects in other samples by inputting a generated image to other samples by a low resolution imaging subsystem to a trained low resolution neural network.