Abstract: In an embodiment an apparatus includes an image acquisition device with at least one camera, the image acquisition device configured to acquire a multi-line section of a recording region and an evaluation device configured to process at least two sub-areas of the multi-line section as one strip image each and compare at least two strip images of a test pattern to each other in a validation mode to check whether deviations of the strip images are detected.

Abstract: A defective picture generation method applied to industrial quality inspection and a defective picture generation apparatus applied to industrial quality inspection are provided. The method includes: acquiring a first workpiece picture set with defects and a second workpiece picture set without defects; determining a defect annotation picture corresponding to each first workpiece picture in the first workpiece picture set; determining a feature value of each second workpiece picture in the second workpiece picture set; training a pix2pixHD network based on the first workpiece pictures, the second workpiece pictures, the defect annotation pictures and the feature values; acquiring a target defect annotation picture from the defect feature database according to the desired defect type; acquiring a target feature value from the picture feature database according to the desired picture type; and inputting the target defect annotation picture and the target feature value into the trained generator.

Abstract: Metrology methods, modules and systems are provided, for using machine learning algorithms to improve the metrology accuracy and the overall process throughput. Methods comprise calculating training data concerning metrology metric(s) from initial metrology measurements, applying machine learning algorithm(s) to the calculated training data to derive an estimation model of the metrology metric(s), deriving measurement data from images of sites on received wafers, and using the estimation model to provide estimations of the metrology metric(s) with respect to the measurement data. While the training data may use two images per site, in operation a single image per site may suffice—reducing the measurement time to less than half the current measurement time. Moreover, confidence score(s) may be derived as an additional metrology and process control, and deep learning may be used to enhance the accuracy and/or speed of the metrology module.

Abstract: Disclosed are a polarizer attachment detection method and device, and a display device. The polarizer attachment detection method includes: controlling an image collection device to collect image data after attachment of a polarizer in a current detection mode; in response to the image data in the current detection mode failing to match prestored standard image data, switching to a next detection mode, controlling the image collection device to collect image data after the attachment of the polarizer in the next detection mode, and marking the image data collected in the next detection mode as new image data; in response to the new image data matching the prestored standard image data, outputting result information that the attachment is correct; and in response to the new image data failing to match the prestored standard image data, outputting result information that the attachment is incorrect.

Abstract: A training system for training a generator neural network arranged to transform measured sensor data into generated sensor data. The generator network is arranged to receive as input sensor data and a transformation goal selected from a plurality of transformation goals and is arranged to transform the sensor data according to the transformation goal.

Abstract: An object of the present invention is to provide an information processing method and a computer program that can suppress an increase in inspection time in the manufacturing process of the monitors. The present invention provides an information processing method comprising: an error calculation step of calculating an error between input image data input to an autoencoder and output image data output from the autoencoder; a similarity calculation step of calculating a similarity between compressed data and reference data based on the compressed data and the reference data, the compressed data being acquired by compressing the input image data in an encoder of the autoencoder; and a determination step of determining whether a display unevenness of the input image data is acceptable based on a relationship between the error and the similarity, the relationship corresponding to a relational expression or a table.

Abstract: A method for training a cell defect detection model includes training a defect classification model that includes an output layer using a plurality of first sample images so that a defect classification model obtained through training is capable of predicting a plurality of first-preset-category defects of a cell, inputting a second sample image to a defect classification model with at least an output layer removed to obtain a sample feature vector of the second sample image, inputting the sample feature vector of the second sample image to a backbone model to obtain a predicted defect classification result of the second sample image, and adjusting, based on a second-preset-category defect and the predicted defect classification result of the second sample image, parameters of the backbone model and the defect classification model with at least the output layer removed.

Abstract: A computer-implemented method and system provides a labelled training dataset. At least one sub-object or component is selected in a CAD model of an object comprising a plurality of sub-objects or components. A plurality of different render images is generated and the different render images contain the at least one selected sub-object or component. The different render images are labelled on the basis of the CAD model to provide a training dataset based on the labelled render images. Also, a computer-implemented method provides a trained function that is trained on the training dataset. A computer-implemented image recognition method uses such the trained function. An image recognition system comprising an image capture device and a data processing system carries out the image recognition method. A computer program comprises instructions that cause the system to carry out the methods.

Abstract: An inspection apparatus and a method of inspecting a semiconductor device are disclosed. The inspection apparatus includes a stage on which a semiconductor device is positioned, a first light source irradiating a high-frequency light onto an inspection area of the semiconductor device to reduce a potential barrier of a PN junction in the semiconductor device, a beam scanner arranged over the semiconductor device and irradiating a charged particle beam onto the inspection area of the semiconductor device to generate secondary electrons, and a defect detector generating a detection image corresponding to the inspection area and detecting, based on a voltage contrast between a reference image and a plurality of detection images, a defect image indicating a defect in the semiconductor device from among the plurality of detection images.

Abstract: Systems and methods for predicting pest susceptibility, comprising steps to receive geocoded geospatial image data of a crop field from sensors, receive microclimate data of the crop field, determine a degree of canopy closure for one or more portions of the crop field, and then generate a pest susceptibility index utilizing a risk model based on the degree of canopy closure and the microclimate data. The pest susceptibility index comprises a measure of a susceptibility of a crop in the crop field to one or more crop pests. In some embodiments, the method also comprises steps to generate a treatment plan (e.g., pesticide application) and to estimate an anticipated return on investment (ROI). The system therefore leverages remote-sensed data, machine data, analytics, and machine learning to enable farmers to predict, prevent, and control the outbreak of crop pests to greatest economic effect. Such a system addresses a fundamental problem in agriculture.

Abstract: Methods and systems for detecting defects on a mask are provided. One method includes generating a database reference image for a multi-die mask by simulation and detecting first defects on the mask by comparing the database reference image to images of the mask generated by an imaging subsystem for a first of the multiple dies. The method also includes generating a die reference image for the first of the multiple dies by applying one or more parameters of the imaging subsystem learned by generating the database reference image to the images generated by the imaging subsystem of one or more of the multiple dies other than the first multiple die. In addition, the method includes detecting second defects on the mask by comparing the die reference image to the images of the mask generated by the imaging subsystem for the first of the multiple dies.

Abstract: A computer implemented method including acquiring a live image of a subject physical sample of a product or material; inputting the live image to a trained generator neural network to generate a defect-free reconstruction of the live image; comparing the defect-free reconstruction of the live image with the live image to determine a difference; and identifying a defect corresponding to the subject physical sample at a location of the determined difference.

Abstract: A system and method for generating a training data set for training a machine learning model to detect defects in specimens is described herein. A computing system cause presentation of an image on a device of a user. The image includes at least one defect on an example specimen. The computing system receives an annotated image from the user. The user annotated the image using an input via the device. The input includes a first indication of a location of the defect and a second indication of a class corresponding to the defect. The computing system adjusts the annotated image to standardize the input based on an error profile of the user and the class corresponding to the defect. The computing system uploads the annotated image for training the machine learning model.

Abstract: Provided is a technique capable of generating a discriminator having high inference precision at low cost. An inspection system according to one aspect of the present invention generates, from first image data, plural pieces of second image data of a product; determines, by a first discriminator, whether to adopt each piece of the second image data as learning data; generates a learning data set by the second image data that is determined to be adopted as learning data; constructs, by applying machine learning using the generated learning data set, a second discriminator that determines the acceptability of a product; and uses the constructed second discriminator to determine the acceptability of a product in target image data.

Abstract: A system and method for determining a mood for a crowd is disclosed. In example embodiments, a method includes identifying an event that includes two or more attendees, receiving at least one indicator representing emotions of attendees, determining a numerical value for each of the indicators, and aggregating the numerical values to determine an aggregate mood of the attendees of the event.

Abstract: An arrangement detector for plate-shaped objects and a lord port including the same are provided. The arrangement detector for plate-shaped objects includes a judgement window setting means, a shape determination means for determining a shape matching rate by superimposing the judgement window on the image captured by the imaging means, and an object judgement means for judging the plate-shaped objects do not exist in the judgement window overlaid on the image captured by the imaging means, in case that the shape matching rate determined by the shape determination means is equal to or less than a predetermined value. A first reference line is a continuous straight line in the judgement window, and a second reference line is a collection of discontinuous line segments linearly arranged in the judgement window.

Abstract: Systems and methods for automated detection of changes in extent of structures using imagery are disclosed, including a non-transitory computer readable medium storing computer executable code that when executed by a processor cause the processor to: align, with an image classifier model, an outline of a structure at a first instance of time to pixels within an image depicting the structure captured at a second instance of time; assess a degree of alignment between the outline and the pixels depicting the structure, so as to classify similarities between the structure depicted within the pixels of the image and the outline using a machine learning model to generate an alignment confidence score; and determine an existence of a change in the structure based upon the alignment confidence score indicating a level of confidence below a predetermined threshold level of confidence that the outline and the pixels within the image are aligned.

Abstract: A radiation image display apparatus that constitutes a radiation imaging system includes a displayer and a hardware processor that acquires image data of a dynamic image constituted of a plurality of frame images, image data of an analysis dynamic image obtained by applying predetermined image processing to the image data of the dynamic image and image data of a related dynamic image which is related to the dynamic image or the analysis dynamic image respectively, and causes the displayer to display the related dynamic image together with the dynamic image and the analysis dynamic image.

Abstract: This disclosure provides systems, methods, and apparatus detecting defects in a substrate. An image of the substrate is compared with a reference image to identify potential defects. Images corresponding to the potential defects are processed sequentially by a set of classifiers to generate a set of images that include a defect. The set of classifiers can be arranged to have increasing accuracy. A subset of the images corresponding to the potential defects is processed by a type classifier that can determine the type, size, and location of the defect in the images. The defects can be further processed to determine the severity of the defects based on the location of the defects on the substrate.

Abstract: Described embodiments relate to a method comprising: determining a candidate document comprising image data and character data and extracting the image data and the character data from the candidate document. The method comprises providing, to an image-based numerical representation generation model, the image data, and generating, by the image-based numerical representation generation model, an image-based numerical representation of the image data. The method comprises providing, to a character-based numerical representation generation model, the character data; and generating, by the character-based numerical representation generation model, a character-based numerical representation of the character data.