Abstract: A system and method for defect detection in a hole array on a substrate is disclosed herein. In one embodiment, a method for defect detection in a hole array on a substrate, includes: scanning a substrate surface using at least one optical detector, generating at least one image of the substrate surface; and analyzing the at least one image to detect defects in the hole array on the substrate surface based on a set of predetermined criteria.

Abstract: An automatic detection method and an automatic detection system for detecting any crack on wafer edges are provided. The automatic detection method includes the following steps. Several wafer images of several wafers are obtained. The wafer images are integrated to create a templet image. Each of the wafer images is compared with the templet image to obtain a differential image. Each of the differential images is binarized. Each of the differential images which are binarized is de-noised. Whether each of the differential images has an edge crack is detected according to pattern of each of the differential images which are de-noised.

Abstract: An apparatus for classifying a defect generated in a substrate, includes: a first storage part for storing a first image data for defect classification determination, which includes a defect region in which the defect is generated and a surrounding region of the defect region; a first estimation part for estimating a first type of defect by using a deep learning system, based on the first image data; a second storage part for storing a second image data for defect classification estimation, which is obtained by expressing the defect region and the surrounding region by a binarized data; a second estimation part for estimating a second type of defect by using a rule-based system, based on an attribute of the defect region extracted from the second image data; and a comprehensive determination part for comprehensively determining a type of defect based on the first and second types of defects.

Abstract: In a real-time traceability method of a width of a defect based on divide-and-conquer provided by the present invention, through the calibration transfer function, the multidimensional eigenvector analysis technology based on the electromagnetic field simulation database of defect scattered dark-field imaging and the adaptive threshold segmentation method, the real-time traceability of the width of the defect greater than and close to the diffraction limit of the system is performed, respectively. The extreme random tree regression model is trained by multidimensional eigenvector analysis technology based on the multidimensional eigenvectors in the electromagnetic field simulation database of the defect scattered dark-field imaging.

Abstract: The present disclosure relates to a method, device, system and computer-program product for setting a lighting condition when an object is checked and a storage medium. The method includes that: the object is lighted by light sources capable of changing lighting parameters, and the object is captured by an image sensor in such lighting parameters to obtain captured images, wherein the object has known label data; and a part of or all of the captured images and the corresponding label data of the object are applied to learning of a machine learning model, and the lighting condition and the check algorithm parameters of the machine learning model is set simultaneously by optimizing both the lighting parameters and the check algorithm parameters, on the basis of a comparison result between an estimation result of the machine learning model and the label data. Therefore, operations are simplified.

Abstract: Methods and systems for selecting defect detection methods for inspection of a specimen are provided. One system includes one or more computer subsystems configured for separating polygons in a care area into initial sub-groups based on a characteristic of the polygons on the specimen and determining a characteristic of noise in output generated by a detector of an inspection subsystem for the polygons in the different initial sub-groups. The computer subsystem(s) are also configured for determining final sub-groups for the polygons by combining any two or more of the different initial sub-groups having substantially the same values of the characteristic of the noise. In addition, the computer subsystem(s) are configured for selecting first and second defect detection methods for application to the output generated by the detector of the inspection subsystem during inspection of the specimen or another specimen.

Abstract: Techniques for time-based memory allocation for a neural network inference are disclosed. A description of a neural network comprising a plurality of operations to be executed across a set of accelerators is received. A plurality of interconnect times at a plurality of partition points within the neural network are calculated. Each of the plurality of interconnect times corresponds to a duration of time for transferring an output feature map from one of the set of accelerators to another of the set of accelerators to be used as an input feature map. A partitioning scheme that divides the plurality of operations into a set of subgraphs is determined based on the plurality of interconnect times. Each of the set of subgraphs is assigned to a different accelerator of the set of accelerators in accordance with the partitioning scheme.

Abstract: The present invention relates to defects inspection on a silicon carbide wafer or an epitaxial layer thereon to determine the location, and adjustment of the location of the scribe line, which is a separation line forming a gap between adjacent dies. The present invention can obtain high efficiency and economy in the semiconductor process using wafers containing various defects in the surface and thin film, by minimizing the effect of wafer defects on the final yield of the semiconductor chip or die, via adjustment of scribe line positions arranged on the wafer.

Abstract: Inspection data that corresponds to potential defects of an object may be received. A first set of locations of first potential defects can be identified. The first set of locations of the first potential defects can be imaged with a review tool to obtain a first set of review images. The first potential defects can be classified based on the first set of review images to obtain first classification results of the first potential defects. An instruction can be determined for the review tool based on the first classification results, the instruction being associated with detecting potential defects. Using the instruction, a second set of locations of second potential defects of the plurality of potential defects to be imaged with the review tool can be identified.

Abstract: According to one embodiment, an inspection apparatus includes an image generation device which generates a second image corresponding to a first image and a defect detection device which detects a defect in the second image with respect to the first image. The defect detection device is configured to extract a first partial region in which an amount of change of a luminance of the first image and an amount of change of a luminance of the second image have a correlation, and correct, in the first partial region, the luminance of the first image with respect to the luminance of the second image.

Abstract: A method for detecting a defect on a substrate, including receiving a first image, generating a second image, by converting the first image to grayscale levels, calculating a first gray level value, having a maximum number of pixels in the second image, and second and third gray level values, having a number of pixels in the second image equal to a predetermined fraction of the first gray level value, from a histogram of the number of pixels respective to the grayscale levels of the second image, converting the second image into a third image having pixels at a level lower than that of the first gray level value and a fourth image having pixels at a level equal to or higher than the first gray level value, generating fifth and sixth images by detecting edges by applying a Canny algorithm to the third and fourth images, respectively.

Abstract: A computer-implemented method includes receiving a neural network model that includes a tensor operation, and dividing the tensor operation into sub-operations. The sub-operations includes at least two sub-operations that have no data dependency between the two sub-operations. The computer-implemented method further includes assigning a first sub-operation in the two sub-operations to a first computing engine, assigning a second sub-operation in the two sub-operations to a second computing engine, and generating instructions for performing, in parallel, the first sub-operation by the first computing engine and the second sub-operation by the second computing engine. An inference is then made based on a result of the first sub-operation, a result of the second sub-operation, or both. The first computing engine and the second computing engine are in a same integrated circuit device or in two different integrated circuit devices.

Abstract: There is provided a system to examine a semiconductor specimen, the system comprising a processor and memory circuitry configured to obtain a training sample comprising an image of a semiconductor specimen and a design image based on design data, train a machine learning module, wherein the training includes minimizing a function representative of a difference between a simulated image generated by the machine learning module based on a given design image, and a corrected image corresponding to a given image after correction of pixel position of the given image in accordance with a given displacement matrix, wherein the minimizing includes optimizing parameters of the machine learning module and of the given displacement matrix, wherein the trained machine learning module is usable to generate a simulated image of a specimen based on a design image of the specimen.

Abstract: An inspection system may receive inspection datasets from a defect inspection system associated with inspection of one or more samples, where an inspection dataset of the plurality of inspection datasets associated with a defect includes values of two or more signal attributes and values of one or more context attributes. An inspection system may further label each of the inspection datasets with a class label based on respective positions of each of the inspection datasets in a signal space defined by the two or more signal attributes, where each class label corresponds to a region of the signal space. An inspection system may further segment the inspection datasets into two or more defect groups by training a classifier with the values of the context attributes and corresponding class labels for the inspection datasets, where the two or more defect groups are identified based on the trained classifier.

Abstract: Systems and methods for in-die metrology using target design patterns are provided. These systems and methods include selecting a target design pattern based on design data representing the design of an integrated circuit, providing design data indicative of the target design pattern to enable design data derived from the target design pattern to be added to second design data, wherein the second design data is based on the first design data. Systems and methods can further include causing structures derived from the second design data to be printed on a wafer, inspecting the structures on the wafer using a charged-particle beam tool, and identifying metrology data or process defects based on the inspection. In some embodiments the systems and methods further include causing the charged-particle beam tool, the second design data, a scanner, or photolithography equipment to be adjusted based on the identified metrology data or process defects.

Abstract: A method of defect discovery can include providing a nuisance bin in a nuisance filter, partitioning the defect population into a defect population partition, segmenting the defect population partition into a defect population segment, selecting from the defect population segment a selected set of defects, computing one or more statistics of the signal attributes of the defects in the defect population segment, replicating the selected set of defects to yield generated defects, shifting the generated defects outside of the defect population segment, creating a training set, and training a binary classifier. This method can be operated on a system. The method can enable a semiconductor manufacturer to determine more accurately the presence of defects that would otherwise have gone unnoticed.

Abstract: In an approach to tracking mass-produced items via a digital twin, one or more computer processors receive one or more images of a product. One or more computer processors retrieve a digital twin template corresponding to the product. One or more computer processors detect an imperfection in the one or more images of the product. One or more computer processors map data associated with the imperfection to the digital twin template. One or more computer processors map data associated with the imperfection to the digital twin template. One or more computer processors assign a unique identifier to the digital twin template that includes the data associated with the imperfection. One or more computer processors store the data associated with the imperfection in association with the unique identifier.

Abstract: This inventions provides an artificial intelligence (A.I.) identified measuring method for a semiconductor image, comprising the steps of: providing an original image of a semiconductor; identifying a type and/or a category of the original image by an artificial intelligence; introducing a predetermined dimension measuring mode corresponding to the identified type and/or the identified category to scan the original image to generate a measurement signal of the original image; and extracting a designated object from the original image to generate a specific physical parameter of the original image after operation based on a measurement signal of the designated object and the measurement signal of the original image.

Abstract: Provided is a method and system for detection of anomalous work pieces that includes computing at least one deviation data signal for a target data signal of a target work piece with respect to reference data signals recorded for a corresponding production process step of a set of reference work pieces, performing a stepwise anomaly detection by data processing of the at least one computed deviation data signal and a process type indicator indicating a type of the production process step using a trained anomaly detection data model to calculate for each time step or path length step of the production process step an anomaly probability that the respective time step or path length step is anomalous, and classifying the target work piece and/or the production process step as being anomalous or not anomalous on the basis of the calculated anomaly probabilities.

Abstract: For the qualification of a mask for microlithography, the effect of an aerial image of the mask on the wafer is ascertained by means of a simulation for predicting the wafer structures producible by means of the mask.

Abstract: Disclosed herein are methods and systems of metadata management for reviewing data from microscopy experimental sessions. Image data from an experimental session is stored in an archive at one or more filepath locations, either locally or on a network. Metadata associated with the image data is stored in a database with a reference to the filepath where the raw image is stored, such that the metadata is associated in the database with the image data. A user can perform post-experimental filtering, sorting, and searching of the underlying image data using the metadata, which allows the image data to be analyzed without duplication of the image data and without manual review of each individual image. The filtered data is presented in an interactive timeline format.

Abstract: Methods of identifying a hot spot from a design layout or of predicting whether a pattern in a design layout is defective, using a machine learning model. An example method disclosed herein includes obtaining sets of one or more characteristics of performance of hot spots, respectively, under a plurality of process conditions, respectively, in a device manufacturing process; determining, for each of the process conditions, for each of the hot spots, based on the one or more characteristics under that process condition, whether that hot spot is defective; obtaining a characteristic of each of the process conditions; obtaining a characteristic of each of the hot spots; and training a machine learning model using a training set including the characteristic of one of the process conditions, the characteristic of one of the hot spots, and whether that hot spot is defective under that process condition.

Abstract: Devices, systems, and methods obtain one or more training images; obtain a test image; select one or more associated pixels in the training images for a target pixel in the training images; calculate respective value relationships between a value of the target pixel and respective values of the associated pixels in the training images; select one or more associated pixels in the test image for a target pixel in the test image; and detect an anomaly in the target pixel in the test image based on the respective value relationships between the value of the target pixel and the respective values of the associated pixels in the training images and on respective value relationships between a value of the target pixel and respective values of the associated pixels in the test image.

Abstract: A processing device includes: a processor including hardware, the processor being configured to acquire image data; determine, for each pixel of an image corresponding to the acquired image data, whether a pixel value of the pixel is equal to or less than a preset threshold as a dark level; accumulate, for a predetermined number of frames, the pixel value that has been determined to be equal to or less than the preset threshold and positional information regarding a position of the pixel whose pixel value has been determined to be equal to or less than the preset threshold, on the image sensor; calculate a statistical value of the accumulated pixel values for each pixel; determine, for each pixel, whether the statistical value falls outside a preset range; and correct a pixel value of the pixel whose statistical value has been determined to fall outside the preset range.

Abstract: Disclosed are semiconductor devices and methods of manufacturing the same. The method comprises providing a layout comprising a first group that includes first and second patterns and a second group that includes third and fourth patterns, examining a bridge risk region in the layout, biasing one end of at least one of the first and third patterns, and forming first to fourth conductive patterns by respectively using the first to fourth patterns of the layout. The one end of at least one of the first and third patterns are adjacent to the bridge risk region.

Abstract: An image processing apparatus includes an acquisition unit which acquires a parallax image generated based on a signal of a photoelectric converter among a plurality of photoelectric converters which receive light beams passing through partial pupil regions of an imaging optical system different from each other, and acquires a captured image generated by combining signals of the plurality of photoelectric converters, and an image processing unit which performs correction process so as to reduce a defect included in the parallax image based on the captured image.

Abstract: A vision inspection system includes a platform supporting parts for inspection at an inspection zone, an inspection station positioned adjacent the platform at the inspection zone including an imaging device to image the parts in a field of view above the upper surface, and a vision inspection controller receiving images from the imaging device. The vision inspection controller includes an auto focus module for orienting the imaging device relative to the inspection zone. The auto focus module determines a working distance for the imaging device from the inspection zone. The auto focus module calculates an image contrast score of pixel values of the images at various working distances from the inspection zone. The vision inspection controller causes the inspection station to operate the imaging device at an imaging working distance corresponding to the working distance associated with the highest image contrast score.

Abstract: Provided is an image processing apparatus including a memory storing at least one instruction, and a processor configured to execute the at least one instruction stored in the memory to obtain first feature information by performing a convolution operation on a first image and a first kernel included in a first convolution layer among a plurality of convolution layers, obtain at least one piece of characteristic information, based on the first feature information; obtain second feature information, based on the first feature information and the at least one piece of characteristic information, obtain third feature information by performing a convolution operation on the obtained second feature information and a second kernel included in a second convolution layer that is a layer next to the first convolution layer among the plurality of convolution layers, and obtain an output image, based on the third feature information.

Abstract: An optical measurement equipment includes an adjustment apparatus and at least two image capturing devices. The image capturing devices have a depth-of-field and attached to the adjustment apparatus. The image capturing devices are adjusted by the adjustment apparatus such that a portion to be measured of a workpiece is located within the depth-of-field of the image capturing devices.

Abstract: An unmanned aerial vehicle (UAV)-based intelligent anomaly identification method for petroleum pipeline inspection is provided. The precise UAV cruise technology is combined with a target detection algorithm to design an intelligent petroleum pipeline inspection method, which realizes fast anomaly detection for petroleum pipeline inspection based on the existing computer processing capability. In addition, a lot of optimization algorithms and improvements are made for target detection under special working conditions of UAV inspection, and a dedicated target detection network model adapted to these conditions is trained. The UAV-based intelligent anomaly identification method realizes accurate, real-time anomaly reporting for petroleum pipeline inspection, reduces the blindness of manual inspection, greatly improves inspection efficiency, reduces labor costs, and has practicability.

Abstract: Testing to detect intermittent electrical pathways is described. Applied currents may be reversed to fully test all components of a workpiece. Various testing methodologies may be employed. These methodologies may include Time Domain Reflectometry (TDR), mechanical agitation, dark current/voltage testing, (dark IV), i.e., electrical testing of a workpiece using applied electricity, and thermographic imaging, e.g., infra-red thermal imaging. The sensed voltage during agitation may be compared to a benchmark voltage to determine whether or not an intermittent failure exists.

Abstract: A method for detecting a screen is provided, which may improve detection accuracy of defective sub-pixels in the display screen. The method includes: obtaining an image of a screen to be detected; performing Gabor filtering on the image of the screen to be detected to obtain a plurality of Gabor filtered images; performing image fusion on the plurality of Gabor filtered images to obtain a fused image; segmenting the fused image by using different gray thresholds to obtain segmented images; and performing defect detection according to the segmented images to determine whether there is a defective sub-pixel in the screen to be detected. A value range of different gray thresholds is within a gray value range of the fused image.

Abstract: Provided is a charged particle beam apparatus capable of estimating an internal device structure of a sample. The charged particle beam apparatus includes an electron beam optical system, a detector, and a calculator. The electron beam optical system irradiates a plurality of irradiation points on a sample, which are different in position or time, with an electron beam. The detector detects electrons emitted from the sample in response to irradiation of the electron beam by the electron beam optical system. The calculator calculates a dependence relationship between the irradiation points based on the electrons detected by the detector at the plurality of irradiation points.

Abstract: Systems and methods for detecting defects on a reticle are provided. One system includes computer subsystem(s) configured for performing at least one repeater defect detection step in front-end processing during an inspection process performed on a wafer having features printed in a lithography process using a reticle. The at least one repeater defect detection step performed in the front-end processing includes identifying any defects detected at corresponding locations in two or more test images by double detection and any defects detected by stacked defect detection as first repeater defect candidates. One or more additional repeater defect detections may be performed on the first repeater defect candidates to generate final repeater defect candidates and identify defects on the reticle from the final repeater defect candidates.

Abstract: Systems are provided for managing defect data objects. A system stores a plurality of defect data objects that have been input to the system, and generates an issue item including one or more defect data objects that are selected from the stored defect data objects based on user input. The system determines similarity between the one or more defect data objects in the issue item and one or more of the stored defect data objects that are out of the issue item, based on comparison of one or more parameter values. The system determines one or more candidate defect data objects to be included in the issue item from the one or more of the stored defect data objects that are out of the issue item based on the similarity, and includes one or more of the determined candidate defect data objects in the issue item based on user input.

Abstract: An industrial image inspection method includes: generating a test latent vector of a test image; measuring a distance between a training latent vector of a normal image and the test latent vector of the test image; and judging whether the test image is normal or defected according to the distance between the training latent vector of the normal image and the test latent vector of the test image.

Abstract: A method for manufacturing an orthodontic aligner includes printing a mold associated with a dental arch of a patient based on a digital model of the mold, forming the orthodontic aligner over the mold, and trimming the orthodontic aligner. The method further includes assessing a quality of the orthodontic aligner by receiving a digital representation of the orthodontic aligner, the digital representation having been generated based on imaging of the orthodontic aligner, analyzing the digital representation of the orthodontic aligner to identify a quality-related property of the orthodontic aligner, determining, based on the quality-related property, that the orthodontic aligner comprises a manufacturing flaw, and classifying the orthodontic aligner as requiring further inspection by a technician based on determining that the orthodontic aligner comprises the manufacturing flaw.

Abstract: Described herein are improvements for identifying defects during automated item production. In one example, a method includes identifying a first defect in a first item. The first defect is associated with a stage of production of the first produced item. The method further includes retrieving first parametric data associated with the stage for the first item and identifying one or more defect indicators based on the first parametric data and second parametric data associated with the stage for one or more second items having defects associated with the stage. The method also includes monitoring subsequent parametric data associated with the stage to recognize the one or more defect indicators in the subsequent parametric data.

Abstract: There is provided with an information processing device. A defect detecting unit detects a defect of an object in an input image. An extracting unit extracts a feature amount pertaining to a partial image of the defect from the input image, on the basis of a result of detecting the defect. An attribute determining unit determines an attribute of the defect using the feature amount pertaining to the partial image of the defect.

Abstract: It is possible to provide an inspection apparatus having a function of automatically checking a validity of an appearance inspection result without manual operation. An inspection apparatus for performing an appearance inspection using a plurality of images obtained by imaging an inspection target includes a determination unit that estimates an inspection result on each of the plurality of images based on a predetermined relationship model between the image and the inspection result and calculates a comprehensive inspection result based on reliability of each of the inspection results on the plurality of images.

Abstract: An initial inspection or critical dimension measurement can be made at various sites on a wafer. The location, design clips, process tool parameters, or other parameters can be used to train a deep learning model. The deep learning model can be validated and these results can be used to retrain the deep learning model. This process can be repeated until the predictions meet a detection accuracy threshold. The deep learning model can be used to predict new probable defect location or critical dimension failure sites.

Abstract: A method and an apparatus for inspecting a corrosion defect of a ladle are provided. The method includes: acquiring images from various angles using an image acquisition apparatus inside a to-be-inspected ladle; and inputting the acquired images into a defect inspection system to obtain a label representing a defect category, the defect inspection system including: a deep convolutional neural network that predicts a category of a corrosion defect of the ladle included in the images based on the input images. This method has the advantages of high safety, high accuracy and high real-time performance in inspecting the ladle status.

Abstract: Provided is a wafer inspection apparatus including a monochromator that extracts monochromatic light, a collimator that outputs the monochromatic light as parallel light, a first polarization assembly that polarizes the parallel light and radiates the polarized light to a wafer, an imaging optical system that condenses light reflected from the wafer, a spectroscope that splits the condensed light into a plurality of spectrums, a first lens that condenses the plurality of spectrums, a second polarization assembly that outputs the plurality of spectrums as a plurality of polarized lights having different diffraction orders and a difference of 90°, a second lens that condenses the plurality of polarized lights, a third polarization assembly that outputs common polarized light based on the plurality of polarized interfering with each other, a camera that generates a phase difference image based on the common polarized light, and a signal processor that analyzes the phase difference image.

Abstract: The invention relates to a system in particular a quantum sensor system, for testing a device-under-test, DUT, comprising: an optically excitable medium which is arranged to receive electromagnetic, EM, radiation emitted by the DUT, at least one light source configured to irradiate the medium with at least one light beam, wherein the medium is optically excited by the at least one light beam, a field generator unit configured to generate an electric and/or magnetic field within the medium, wherein a resonance frequency of the excited medium is modified by an amplitude of the electric and/or magnetic field, wherein an optical parameter, in particular a luminescence, of the exited medium is locally modified if a frequency of the EM radiation corresponds to the resonance frequency at a position in the medium, an image detector configured to acquire an image of the medium, wherein the image shows an intensity profile that results from the modification of the optical parameter, a processor configured to analyze th

Abstract: A method for correcting a processing condition includes imaging a substrate using an imaging device before start and after completion of a series of processings; specifying a processing apparatus estimated as having an abnormality, based on an imaging result and information on the processing apparatus; performing the unit processing in the processing apparatus on an inspection substrate under a predetermined processing condition, and imaging the inspection substrate by the imaging device before and after performing the unit processing; determining presence/absence of an actual abnormality in the processing apparatus specified in the specifying; and correcting, with respect to the processing apparatus determined as having the actual abnormality in the determining presence/absence of an actual abnormality, the processing condition of the unit processing in the processing apparatus based on the imaging result in the imaging the inspection substrate for determining an abnormality.

Abstract: Some embodiments of devices, systems, and methods generate a respective connection mask at each defect detection location in a binary defect map, wherein each detection location in the binary defect map has a value that indicates that the detection location either has a defect or, alternatively, does not have a defect; generate one or more clusters of defect detection locations in the binary defect map based on each defect location's respective connection mask and on the binary defect map; generate respective bounding boxes for the one or more clusters; and recursively perform the following: checking respective sizes of the bounding boxes, and splitting any clusters of the one or more clusters that have respective bounding boxes with sizes that exceed one or more thresholds, thereby generating additional clusters of defect locations.

Abstract: An image processing apparatus includes an identifiability value obtaining portion, an inspection region generation portion, and an image inspection portion. The identifiability value obtaining portion is configured to obtain, for each pixel address constituting an image plane, an identifiability value for identifying which of a first inspection result and a second inspection result the pixel address corresponds to. The inspection region generation portion is configured to generate an inspection region serving as a target of image processing by setting a portion of the image plane including the pixel address where the obtained identifiability value satisfies a specific condition as the inspection region. The image inspection portion is configured to perform image processing for inspection on a partial image corresponding to the inspection region among a third image obtained by imaging a third target object.

Abstract: A measurement system is provided, including a measurement machine and a computer. The measurement machine is configured to measure a thickness T1 of a to-be-tested circuit board and a drilling depth D1 of the to-be-tested circuit board. The computer calculates a length S1 of a residual conductive portion in a back drilled hole of the to-be-tested circuit board according to a thickness T of a reference circuit board, a drilling depth D of the reference circuit board, a length S of a residual conductive portion in a back drilled hole of the reference circuit board, the thickness T1 of the to-be-tested circuit board and the drilling depth D1 of the to-be-tested circuit board.

Abstract: Defects from a hot scan can be saved, such as on persistent storage, random access memory, or a split database. The persistent storage can be patch-based virtual inspector virtual analyzer (VIVA) or local storage. Repeater defect detection jobs can determined and the wafer can be inspected based on the repeater defect detection jobs. Repeater defects can be analyzed and corresponding defect records to the repeater defects can be read from the persistent storage. These results may be returned to the high level defect detection controller.

Abstract: An image generation apparatus configured to generate a substrate image for inspection regarding a defect on a substrate, the substrate having a frame pattern formed on a surface thereof, the frame pattern being a unique pattern for each kind of a treatment recipe for the substrate, the image generation apparatus including: a region estimator configured to estimate a region corresponding to the frame pattern in a substrate image of an inspection object based on an identification model, the identification model being acquired by machine learning in advance and for identifying an image of the frame pattern included in a substrate image; and an eraser configured to erase the image of the frame pattern from the substrate image of the inspection object based on an estimation result by the region estimator to generate the substrate image for inspection.