Abstract: A system for image-based validation of a target object relative to a 3D control model is configured to retrieve a 3D control model, identify coordinates for each of the components, determine virtual camera locations using a camera placement algorithm to enable a multi-view image capture of all the components of the 3D control model, position virtual cameras at the determined camera locations, capture, by real-life cameras positioned at real-life camera locations respectively corresponding to the virtual camera locations, a 3D target model of the target object, receive the 3D target model captured by the real-life cameras, generate 2D target planar images of the target object, generate 2D control planar images of the control object, detect at least one difference between the 2D target planar images and the 2D control planar images, generate a validation output indicating the at least one difference, and output the validation output.

Abstract: A system for automated surface anomaly detection includes at least one processor which is configured to retrieve from non-volatile memory a 3D control image; capture a 3D target image depicting at least the portion of the exterior of a target object; generate 2D target planar images of the target object based on the 3D target image, using a first plurality of virtual cameras; generate 2D control planar images of the control object based on the 3D control image using a second plurality of virtual cameras, the 2D control planar images corresponding to the 2D target planar images of the target object; detect at least one difference between the 2D target planar images and the 2D control planar images; and generate and cause to be displayed an output image comprising a depiction of the target object with the at least one difference indicated.

Abstract: Techniques for adversarial attack avoidance for machine learning (ML) are disclosed. These techniques include receiving one or more images at a trained ML model and receiving attack data at the ML model. The techniques further include predicting an object depicted in the one or more images using the ML model, based on the one or more images, metadata relating to the one or more images, and the attack data. The ML model uses the metadata to prevent the attack data from changing a result of the predicting.

Abstract: A computing system is configured to collect input data related to at least a portion of an object or an environment from the input sensor, execute a feature extractor to extract features for data elements of the input data, execute a clustering model configured to cluster the data elements of the input data into feature clusters based on similarities of the extracted features to each other, label a target clusters of the feature clusters and data elements of the target clusters with respective predetermined labels, generate a training dataset including the data elements of the target clusters, and train a machine learning model using the training dataset to predict a label for an inference time input data element at inference time. The respective predetermined labels of the target clusters correspond to prediction labels of the machine learning model.

Abstract: A computing device, a method and a computer program product are provided to generate realistic synthetic image data that is utilized to train, test and/or deploy a computer vision model, such as to recognize an object from an image, to identify a visual anomaly in the object from an image of the object or the like. In the context of a method, a plurality of computer generated three-dimensional (3D) models of an object are generated. The method also generates realistic synthetic image data representative of each of the plurality of computer generated 3D models. The realistic synthetic image data is generated by utilizing a first machine learning model trained with pairs of computer generated synthetic image data and real image data of the object. The method further trains, tests or deploys the computer vision model to recognize the object from an image based upon the realistic synthetic image data.

Abstract: A method, apparatus, system, and computer program product for training a gesture recognition machine learning model system. Temporal images for a set of gestures used for ground operations for an aircraft are identified by a computer system. Pixel variation data identifying movement on a per image basis from the temporal images is generated by the computer system. The temporal images and the pixel variation data form training data. A set of feature machine learning models is trained by the computer system to recognize features using the training data.

Abstract: An image generation system is provided to: receive a 3D CAD (computer aided design) model comprising 3D model images of a target object; based on the 3D CAD model of the target object, generate augmented CAD models of the target object comprising data sets, each data set respectively corresponding to an associated one of a plurality of attribute classes, each data set comprising a plurality of 2D model images; input the data sets into a generative adversarial network comprising generators respectively corresponding to the plurality of attribute classes and discriminators respectively corresponding to the generators; generate synthetic photorealistic images of the target object using the generators, the synthetic photorealistic images including attributes in accordance with the data sets corresponding to the plurality of attribute classes; and output the synthetic photorealistic images of the target object.

Abstract: A human machine interface (HMI), a method of interacting with an HMI and a corresponding computer program product facilitate user interaction with an HMI. In a method, a face of a person attempting to interact with the HMI is detected and a gesture made by the person relative to the HMI is also detected. The method analyzes information regarding the face of the person that has been detected to determine whether the person is wearing personal protection equipment. In an instance in which the person is determined to not be wearing personal protection equipment, the method suspends a response to the gesture made by the person. However, in an instance in which the person is determined to be wearing personal protection equipment, the method analyzes the gesture made by the person relative to the HMI and converts the gesture to a corresponding command to the system associated with the HMI.

Abstract: Certain aspects of the present disclosure provide techniques for autonomous image acquisition. This includes determining a plurality of two-dimensional image perspectives for a plurality of image capture devices, and comparing the plurality of two-dimensional image perspectives with a generated two-dimensional representation of a target object, where the two-dimensional representation is generated based on a three-dimensional model of the target object. This further includes automatically moving at least one of the plurality of image capture devices, based on the comparing, to increase a portion of the target object captured by the plurality of image capture devices.

Abstract: An apparatus for inspecting a surface comprises a light source and a camera. When a light beam is generated by the light source, a centerline of the light beam is coincident with a line of sight of the camera when viewed in a direction, perpendicular to a first plane that contains one of the centerline of the light beam or the line of sight of the camera. The centerline of the light beam is parallel to the line of sight of the camera when viewed in a direction, perpendicular to a second plane that is perpendicular to the first plane and that contains the centerline of the light beam and the line of sight of the camera.

Abstract: An apparatus for inspecting a surface comprises a light source and a camera. When a light beam is generated by the light source, a centerline of the light beam is coincident with a line of sight of the camera when viewed in a direction, perpendicular to a first plane that contains one of the centerline of the light beam or the line of sight of the camera. The centerline of the light beam is parallel to the line of sight of the camera when viewed in a direction, perpendicular to a second plane that is perpendicular to the first plane and that contains the centerline of the light beam and the line of sight of the camera.

Abstract: An image generation system is provided to: receive a 3D CAD (computer aided design) model comprising 3D model images of a target object; generate a graph data structure; based on the 3D CAD model of the target object and the graph data structure, generate a plurality of augmented CAD models of the target object comprising a plurality of data sets, each data set respectively corresponding to an associated one of a plurality of attribute classes, each data set comprising a plurality of 2D model images; input the plurality of data sets into a generative adversarial network; generate synthetic photorealistic images of the target object using the plurality of generators of the generative adversarial network, the synthetic photorealistic images including attributes in accordance with the plurality of data sets corresponding to the plurality of attribute classes; and output the synthetic photorealistic images of the target object.

Abstract: Systems and methods for detecting surface anomalies are disclosed. For example, a computer-implemented method for detecting surface anomalies on an object comprises receiving measured electromagnetic radiation (EMR) profiles for the object, generating synthetic EMR profiles for the object based on the measured EMR profiles, determining whether the object contains a surface anomaly based on the measured EMR profiles and the synthetic EMR profiles, and indicating a surface anomaly to a user via a display when a surface anomaly is detected. In another example, a system comprises a computing device comprising non-transitory memory with computer-readable instructions for receiving unpaired image data of an object of two different image types, predicting missing image data to generate paired image data of the two different image types, and determining whether the object contains a surface anomaly based on the paired image data.

Abstract: A computer-implemented method for extracting information from a diagram comprises identifying, via first machine learning logic implemented by a computer, features associated with an article illustrated in the diagram. A knowledge graph that defines relationships between the features is generated via second machine learning logic implemented by the computer and based on the features. The knowledge graph specifies nodes associated with the features and edges between the nodes that specify the relationships between the features. A request for information associated with the diagram is received by the computer and from a user. The computer searches the knowledge graph for the information associated with the request and communicates the information associated with the request to the user.

Abstract: An image generation system is provided to: receive a 3D CAD (computer aided design) model comprising 3D model images of a target object; based on the 3D CAD model of the target object, generate augmented CAD models of the target object comprising data sets, each data set respectively corresponding to an associated one of a plurality of attribute classes, each data set comprising a plurality of 2D model images; input the data sets into a generative adversarial network comprising generators respectively corresponding to the plurality of attribute classes and discriminators respectively corresponding to the generators; generate synthetic photorealistic images of the target object using the generators, the synthetic photorealistic images including attributes in accordance with the data sets corresponding to the plurality of attribute classes; and output the synthetic photorealistic images of the target object.

Abstract: An image generation system is provided to: receive a 3D CAD (computer aided design) model comprising 3D model images of a target object; generate a graph data structure; based on the 3D CAD model of the target object and the graph data structure, generate a plurality of augmented CAD models of the target object comprising a plurality of data sets, each data set respectively corresponding to an associated one of a plurality of attribute classes, each data set comprising a plurality of 2D model images; input the plurality of data sets into a generative adversarial network; generate synthetic photorealistic images of the target object using the plurality of generators of the generative adversarial network, the synthetic photorealistic images including attributes in accordance with the plurality of data sets corresponding to the plurality of attribute classes; and output the synthetic photorealistic images of the target object.

Abstract: A method for managing nonconformances in laminates. The method comprises recording, by a sensor system, layup information about a layup of layers on a workpiece platform, wherein the layup of layers forms a workpiece and recording inspection information about the laminate on an inspection platform, wherein the laminate is formed from curing the workpiece. An analyzer in a computer system identifies a laminate nonconformance in the laminate using the inspection information and a user input verifies the laminate nonconformance in the laminate is present. An artificial intelligence system is trained by the computer system using the layup information, the inspection information, and the user input verifying the laminate nonconformance.

Abstract: A computing device, a method and a computer program product are provided to generate realistic synthetic image data that is utilized to train, test and/or deploy a computer vision model, such as to recognize an object from an image, to identify a visual anomaly in the object from an image of the object or the like. In the context of a method, a plurality of computer generated three-dimensional (3D) models of an object are generated. The method also generates realistic synthetic image data representative of each of the plurality of computer generated 3D models. The realistic synthetic image data is generated by utilizing a first machine learning model trained with pairs of computer generated synthetic image data and real image data of the object. The method further trains, tests or deploys the computer vision model to recognize the object from an image based upon the realistic synthetic image data.

Abstract: A system for automated surface anomaly detection includes at least one processor which is configured to retrieve from non-volatile memory a 3D control image; capture a 3D target image depicting at least the portion of the exterior of a target object; generate 2D target planar images of the target object based on the 3D target image, using a first plurality of virtual cameras; generate 2D control planar images of the control object based on the 3D control image using a second plurality of virtual cameras, the 2D control planar images corresponding to the 2D target planar images of the target object; detect at least one difference between the 2D target planar images and the 2D control planar images; and generate and cause to be displayed an output image comprising a depiction of the target object with the at least one difference indicated.

Abstract: A human machine interface (HMI), a method of interacting with an HMI and a corresponding computer program product facilitate user interaction with an HMI. In a method, a face of a person attempting to interact with the HMI is detected and a gesture made by the person relative to the HMI is also detected. The method analyzes information regarding the face of the person that has been detected to determine whether the person is wearing personal protection equipment. In an instance in which the person is determined to not be wearing personal protection equipment, the method suspends a response to the gesture made by the person. However, in an instance in which the person is determined to be wearing personal protection equipment, the method analyzes the gesture made by the person relative to the HMI and converts the gesture to a corresponding command to the system associated with the HMI.