A method and apparatus for identifying locations to be inspected on a substrate is disclosed. A defect location prediction model is trained using a training dataset associated with other substrates to generate a prediction of defect or non-defect and a confidence score associated with the prediction for each of the locations based on process-related data associated with the substrates. Those of the locations determined by the defect location prediction model as having confidences scores satisfying a confidence threshold are added to a set of locations to be inspected by an inspection system. After the set of locations are inspected, the inspection results data is obtained, and the defect location prediction model is incrementally trained by using the inspection results data and process-related data for the set of locations as training data.

A method of performing 3D metrology on a structure includes directing an electron beam onto a surface of the structure, capturing a first set of images of the structure at a first landing angle, capturing a second set of images of the structure at a second landing angle, the second landing angle being different from the first landing angle, and determining, by comparing the first set of images to the second set of images, at least one 3D parameter of the structure. The first set of images at the first landing angle and the second set of images at the second landing angle are captured in a single run of the electron beam across the surface of the structure.