Abstract: A system and method for path navigation accuracy estimation. A method includes determining a degree of obstruction for each of a plurality of objects shown in visual content, wherein the degree of obstruction for each object represents a degree to which the object will obstruct visibility by a vehicle capturing visual content in an environment where the vehicle is navigating and the plurality of objects are disposed; determining, for each of a plurality of operational points, a co-visibility of the vehicle at the operational point based on a pose of the vehicle at the operational point and the degree of obstruction for each of the objects which is within view of the vehicle at the operational point; and estimating a visual navigation performance at each of the operational points based on the co-visibility determined for the operational point.

Abstract: Systems and methods of monitoring inventory of a product storage facility include an image capture device configured to move about the product storage areas of the product storage facility and capture images of the product storage areas from various angles. A computing device coupled to the image capture device obtains the images of the product storage areas captured by the image capture device and processes the obtained images of the product storage areas to detect individual products captured in the obtained images. Based on detection of the individual products captured in the images, the computing device analyzes each of the obtained images to extract meta data from the packaging the individual products to detect one more keywords and determine the locations of the detected keywords on the packaging, and then utilize this information to predict an identity of the products associated with the packaging.

Abstract: Techniques are described for a model-based process to validate build integrity of software products. A build integrity validation system generates a build artifact model including one or more software components extracted from a build artifact resulting from a software build process of source code and generates a source code including one or more software components extracted from the source code. The system compares the software components of the build artifact model and the software components of the source code model, and identifies at least one discrepancy between the build artifact model and the source code model. The system determines a level of risk associated with the at least one discrepancy and, in response to determining the level of risk associated with the at the least one discrepancy, generates a decision indicating whether the at least one build artifact is approved for use in a software product.

Abstract: Disclosed is a computing device for generating region information of at least one object included in an image, the computing device including: a memory including computer-executable components; and a processor for executing the computer-executable components stored in the memory, in which the computer-executable components include: a key point heat map generation component for generating, for the image, a key point heat map including key point information of the at least one object; a non-rotating bounding box generation component for generating a non-rotating bounding box based on the generated key point heat map; a rotating bounding box generation component for generating a rotating bounding box by using the non-rotating bounding box; and a final bounding box generation component for representing a region occupied by the at least one object in the image by using at least one of the non-rotating bounding box or the rotating bounding box.

Abstract: A system includes a first information processing apparatus and a second information processing apparatus capable of communicating with each other. The first information processing apparatus is configured to acquire three-dimensional point cloud data including multiple vertices each having three parameters of x-, y-, and z-coordinates. A combination of the x- and y-coordinates is different from vertex to vertex. The first or second information processing apparatus is configured to convert the three-dimensional point cloud data into two-dimensional point cloud data in which each of the multiple vertices has two parameters of a parameter m with a different value for each combination of the x- and y-coordinates and the z-coordinate. The second information processing apparatus is configured to generate mesh data, using the two-dimensional point cloud data and information indicating correspondence between the combination of the x- and y-coordinates and the parameter m.

Abstract: A virtual try-on system includes one or more hardware processors configured to function as a learning unit, an acquisition unit, a deriving unit, and a generation unit. The learning unit learns, by machine learning using three-dimensional data of a try-on person, a learning model using a teacher try-on person image as input data, and using a body shape parameter indicating a body shape of the try-on person represented by the teacher try-on person image and compositing position information of a clothing image in the teacher try-on person image as output data. The acquisition unit acquires a try-on person image. The deriving unit derives output data of the try-on person represented by the try-on person image using the try-on person image and the learning model. The generation unit generates a virtual try-on image by compositing the try-on person image and the clothing image using the derived output data.

Abstract: The present invention is applicable to the fields of panoramic images and videos, and provides a panoramic image and video splicing method, and a panoramic camera. In the present invention, fisheye photos captured by two adjacent cameras are mapped to corresponding seam areas of a sphere model, so as to form two strip graphs having overlapping areas; performing block template matching on the two strip graphs to obtain an initial template matching result; performing matching filtering on the initial template matching result by using an area expansion-based matching filtering algorithm, so as to obtain a final matching result; and updating a mapping relationship between the fisheye photos and the corresponding seam areas of the sphere model, and performing panoramic splicing according to the updated mapping relationship to obtain a seamless panoramic image.

Abstract: A device and a method for controlling one or more robots. The method includes: for each robot of the one or more robots, determining a labeled Markov decision process for each skill of one or more skills which the robot is able to implement, each labeled Markov decision process having a labeling function which indicates for each state of the labeled Markov decision process, whether one or more control conditions of a plurality of predetermined control conditions is/are satisfied; providing a control mission defined by a mission specification, the mission specification having a time sequence over a subset of the plurality of predetermined control conditions; and controlling the one or more robots to execute the control mission in such a way that the control conditions of states of the one or more robots contained in the mission specification are satisfied.

Abstract: Provided is an information processing device capable of determining accuracy of data acquired using an LIDAR sensor. An information processing device (10) according to the present disclosure includes: an image-capturing sensor (11); a learning unit (12) configured to learn, as learning data, learning image-capturing data and likelihood of a distance between points included in 3D point cloud data with correct answer in a region substantially the same region included in the learning image-capturing data, and to generate a learned model; and an estimation unit (13) configured to use the learned model to generate estimation data including likelihood of a distance between points included in estimation 3D point cloud data determined based on estimation image-capturing data, from the estimation image-capturing data acquired by the image-capturing sensor (11).

Abstract: A method for teaching and controlling a robot to perform an operation based on human demonstration with images from a camera. The method includes a demonstration phase where a camera detects a human hand grasping and moving a workpiece to define a rough trajectory of the robotic movement of the workpiece. Line features or other geometric features on the workpiece collected during the demonstration phase are used in an image-based visual servoing (IBVS) approach which refines a final placement position of the workpiece, where the IBVS control takes over the workpiece placement during the final approach by the robot. Moving object detection is used for automatically localizing both object and hand position in 2D image space, and then identifying line features on the workpiece by removing line features belonging to the hand using hand keypoint detection.

Abstract: A method and an apparatus for controlling cruise of an unmanned air vehicle based on a prefabricated construction platform, a terminal device, and a computer readable storage medium are provided. The control method comprises: acquiring personnel information at the construction site; adjusting a cruise cycle of the unmanned air vehicle based on the personnel information; transmitting the imagery data of the construction site collected by the unmanned air vehicle to the prefabricated construction platform; and completing three-dimensional modeling on the prefabricated construction platform based on the imagery data, and displaying a resulting model. In this way, the technical problem that the existing cruise method of the unmanned air vehicle has a low level of intelligence is solved.

Abstract: The present invention provides an image processing method, wherein the image processing method includes the steps of: using a plurality of corner detection filters to perform corner detection on a specific pixel of image data to generate a plurality of detection results, respectively, wherein the plurality of corner detection filters correspond to a plurality of corners with different directions, respectively; determining which one of the plurality of corners with different directions the specific pixel belongs to according to the plurality of detection results; and according to a specific corner among the plurality of corners to which the specific pixel belongs, performing an image processing operation corresponding to the specific corner on the specific pixel to generate processed image data.

Abstract: The present invention relates to a method and an apparatus for composing a background and a face by using a deep learning network, comprising: receiving an input of an original face image and a converted face image, and extracting data preprocessing and feature vectors for each image; generating a face feature vector mask from the extracted feature vectors; and generating a composite image by performing adaptive object normalization on the basis of the generated face feature vector mask.

Abstract: Embodiments relate to a method, an electronic device, and a program product for training an encoder and processing data. The method includes inputting sample point cloud data for an object to an encoder to obtain encoded data for the object, and determining, by transforming the encoded data, a plurality of invariant portions for the object and a plurality of variable portions for the object, an invariant portion in the plurality of invariant portions indicating an invariant feature of the object and a variable portion in the plurality of variable portions indicating a variable feature of the object. The method further includes determining, based on the plurality of invariant portions and the plurality of variable portions, a similarity loss and a spatial loss for the sample point cloud data, and adjusting, based on the similarity loss and the spatial loss, a parameter of the encoder to obtain a trained encoder.

Abstract: The present disclosure provides a method and apparatus for adjusting a virtual face model, an electronic device and a storage medium, so as at least to solve the technical problems of tedious operation steps and poor user experience of conventional solutions for adjusting a virtual image. A specific implementation solution includes: a real human face in a reality scene is identified to obtain a first virtual face model of the real human face in a virtual scene; in response to a face-pinching operation acting on the real human face, the face-pinching operation is identified to determine at least one to-be-adjusted key point in the first virtual face model and an adjustment manner of the at least one to-be-adjusted key point; and the first virtual face model is adjusted to a second virtual face model based on the adjustment manner of the at least one to-be-adjusted key point.

Abstract: The disclosure provides a classification method and an electronic apparatus. The classification method includes the following steps. First feature data of multiple pictures of assembly is extracted, and each picture of assembly includes an operator at a station. The first feature data is converted into a first feature vector. Second feature data recording personal data of the operator is converted into a second feature vector. The first feature vector and the second feature vector are merged into a first feature matrix. The efficiency of the operator operating at the station is classified according to the first feature matrix to obtain a classification result.

Abstract: An exemplary processing system accesses a three-dimensional (3D) model of an anatomical structure of a patient and applies a detection process to the 3D model to detect a single-layer anatomical feature in the anatomical structure. The detection process includes generating, from the 3D model, a probability map of candidate points for the single-layer anatomical feature, and generating, based on the probability map of candidate points, a single-layer mesh representing the single-layer anatomical feature.

Abstract: A Neonatal CNN (N-CNN) is provided for detecting neonatal pain emotion based upon facial recognition. A cascaded N-CNN is trained using a Neonatal Pain Assessment Database (NPAD) to automatically identify a neonatal patient experience pain in real-time. These results show that the automatic recognition of neonatal pain provided by the embodiments of the present invention is a viable and more efficient alternative to the current standard of pain assessment.

Abstract: A system and method for performing object detection are presented. The system receives spatial structure information associated with an object which is or has been in a camera field of view of a spatial structure sensing camera. The spatial structure information is generated by the spatial structure sensing camera, and includes depth information for an environment in the camera field of view. The system determines a container pose based on the spatial structure information, wherein the container pose is for describing at least one of an orientation for the container or a depth value for at least a portion of the container. The system further determines an object pose based on the container pose, wherein the object pose is for describing at least one of an orientation for the object or a depth value for at least a portion of the object.

Abstract: A three-dimensional position and posture recognition device speeds estimation of a position posture and a gripping coordinate posture of a gripping target product. The device includes: a sensor unit configured to measure a distance between an image of an object and the object; and a processing unit configured to calculate an object type included in the image, read model data of each object from the external memory, and create structured model data having a resolution set for each object from the model data, generate measurement point cloud data of a plurality of resolutions from information on a distance between an image of the object and the object, perform a K neighborhood point search using the structured model data and the measurement point cloud data, and perform three-dimensional position recognition processing of the object by rotation and translation estimation regarding a point obtained from the K neighborhood point search.

Abstract: To provide a lane shape recognition system and a lane shape recognition method which can determine an effective range of the approximated curve corresponding to the lane shape. A lane shape recognition system and a lane shape recognition method detects relative positions of a series of marks or objects which are continuously arranged in front of an own vehicle, with respect to the own vehicle; calculates a curve which approximates the relative positions of the series of marks or objects with respect to the own vehicle; and when a plurality of curves are calculated, mutually compare shapes of plurality of curves and determines the curves corresponding to the lane shape and the effective range of the curves.

Abstract: A method of localizing a device within an environment includes a step, in which a first location information is acquired that is representative of a first specified position. In another step, a first sensor information is acquired that is representative of the first specified position. In another step, the first sensor information is allocated to the first location information in order to generate a first condition information that is unique for the first specified position. In another step, a knowledge base is provided which includes the first condition information. In another step, the device provides a current sensor information indicating a current condition of the environment present at the current position of the device. In another step, the device is localized based on a matching between the current sensor information and the plurality of condition information.

Abstract: Technologies for performing sensor fusion include a compute device. The compute device includes circuitry configured to obtain detection data indicative of objects detected by each of multiple sensors of a host system. The detection data includes camera detection data indicative of a two or three dimensional image of detected objects and lidar detection data indicative of depths of detected objects. The circuitry is also configured to merge the detection data from the multiple sensors to define final bounding shapes for the objects.

Abstract: A device for detection of surface defects on a terminal surface of an optical fiber. The device includes a digital microscope configured to capture an image of a terminal surface; and a mechanism for analyzing the image configured to detect surface defects present on the terminal surface, the analysis mechanism integrating a “U-Net”-type neural network having had its training phase carried out via an enhancer. The enhancer is configured to create training images, intended to train the neural network, based on reference images. The training images are obtained from the reference images by only applying flips, rotations and/or luminosity, contrast, or shade variations to the reference images.

Abstract: Disclosed is a method for recognizing a bitting code of a key. The method comprises: using a back light source and/or a lateral light source to collect an image of a key to be recognized; and recognizing, on the basis of the image, a bitting code of the key to be recognized. Further disclosed are an apparatus for recognizing a bitting code of a key, and a storage medium and an image collection device. The present disclosure can improve the efficiency of recognizing a bitting code of a key and reduce the wear of the key during a recognition process.

Abstract: In the field of robot technologies, a robot control method and apparatus are for protecting safety of a human during interaction between the human and a robot. The method includes: detecting a current location of the human; determining at least two regions surrounding the human according to the detected current location; and controlling movement of the robot in any region of the at least two regions, to protect safety of the human during interaction with the robot. Because the region is set surrounding the human, movement of the human does not affect the interaction between the human and the robot. In addition, using a protected object as a target, the robot is controlled to move in regions that surround the human. Compared with a case in which the robot is limited in fixed space, no matter how the human moves, safety of the human may be effectively protected.

Abstract: An image processing apparatus includes at least one processor executing the following processing including: image acquiring processing that acquires an image of an object in which a plurality of subjects, which are individually identifiable, are provided in a predetermined direction, the image including at least two subjects among the plurality of subjects; identifying processing that generates an information image including a feature amount of each of the subjects included in the image acquired in the image acquiring processing, and identifies each of the subjects by comparing with an information image stored in advance in memory; and output control processing that measures a distance between the subjects, each identified in the identifying processing, and controls to output a measurement result.

Abstract: Glyph accessibility techniques are described. In an example, the glyph management system obtains alternate glyphs of glyphs from an input glyph sequence. The glyph management system then determines a swash type of swashes included in respective alternate glyphs. The swash type is used to generate candidate glyph sequences using one or more rules. The candidate glyph sequences are usable to replace an input glyph sequence in the user interface.

Abstract: A system and method for scoring trained probes for use in analyzing one or more candidate poses of a runtime image is provided. A set of probes with location and gradient direction based on a trained model are applied to one or more candidate poses based upon a runtime image. The applied probes each respectively include a discrete set of position offsets with respect to the gradient direction thereof. A match score is computed for each of the probes, which includes estimating a best match position for each of the probes respectively relative to one of the offsets thereof, and generating a set of individual probe scores for each of the probes, respectively at the estimated best match position.

Abstract: This disclosure relates to method and system for visual inspection of rotating components. The method includes representing rotation cycles of a rotating component as spatial features based on video or image frames, ascertaining and/or evolving Hidden Markov Model (HMM) chains for the cycles, ascertaining a count of the rotating component in the frames and/or labelling the frames with ascertained states of the HMM chains.

Abstract: A simulation system comprising: a plurality of physical components, each corresponding to one of a plurality of physical component types, an attachment panel comprising an arrangement of attachment locations, such that one or more of the physical components are attachable to the attachment panel; a display system configured to provide a visualisation on or proximate one or both of the attachment panel and the plurality of physical components; a capture device configured to capture image data of a current status of the attachment panel and the plurality of physical components; and a controller, when at least one physical component is coupled to the attachment panel.

Abstract: Vector object stylization techniques from raster objects are described that support editing of vector objects in a manner that maintains an underlying mathematical representation of object. A raster object, for instance, is generated from an edited version of an output of a vector object. This raster object, along with the vector object are received as inputs by a vector conversion system. These inputs are utilized by the vector conversion system to generate a stylized vector object having a visual appearance that mimics and simulates a visual appearance of the raster object. As a result, the stylized vector object provides a mathematical representation of the raster object.

Abstract: A method for detecting a change in the rolling radius of a tire and a method for monitoring the state of the tire are based on the analysis of signals representing an actual speed of a vehicle on which the tyre is fitted and a speed of rotation of the wheel bearing the vehicle.

Abstract: An apparatus for illuminating objects nay include: an imaging device configured to capture an image of an area and to generate a first signal representative of the captured image, wherein the imaging device is further configured to be activated to capture the image of the area under uniform illumination; a processing unit configured to receive the first signal; process the first signal to identify one or more shapes bounded by edges in an acquired image; and process the acquired image by filling each of the identified one or more shapes with a first color, and by filling parts of the acquired it age external to the edges with a second color, to generate a second signal representative of the processed image; and an image projection device configured to receive the second signal; generate a light beam representative of the processed image; and direct the light beam to the area.

Abstract: A modeling support system stores a combination of a CAD model and a mesh model corresponding to the CAD model; recognizes a shape indicated by the CAD model; and classifies respective parts of the stored CAD model into one of a plurality of groups. The modeling support system also specifies component parts of an input CAD model and specifies groups corresponding to the specified component parts from the specified groups. A similar-shape-model search unit searches and obtains the part of the CAD model having a shape most similar to the shape indicated by the specified component part, from the part of the CAD model in the specified groups. Further, a mesh model generating unit obtains a mesh model corresponding to the obtained part of the CAD model, from the stored mesh model, and generates a new mesh model.

Abstract: An image processing apparatus comprises a changing unit configured to change a display area of an image from a first display area to a second display area including at least a portion of the first display area, an acquiring unit configured to acquire a first value indicating luminance, in which brightness contrast is considered, in an image displayed in the first display area and a second value indicating luminance, in which brightness contrast is considered, in an image displayed in the second display area, and a correcting unit configured to correct luminance of the image displayed in the second display area based on the first value and the second value that are acquired by the acquiring unit.

Abstract: Systems and methods are disclosed to identify a presence of a volumetric medium in an environment associated with a LIDAR system. In some implementations, the LIDAR system may emit a light pulse into the environment, receive a return light pulse corresponding to reflection of the emitted light pulse by a surface in the environment, and determine a pulse width of the received light pulse. The LIDAR system may compare the determined pulse width with a reference pulse width, and determine an amount of pulse elongation of the received light pulse. The LIDAR system may classify the surface as either an object to be avoided by a vehicle or as air particulates associated with the volumetric medium based, at least in part, on the determined amount of pulse elongation.

Abstract: Systems and methods for estimating depth from projected texture using camera arrays are described. A camera array includes a conventional camera and at least one two-dimensional array of cameras, where the conventional camera has a higher resolution than the cameras in the at least one two-dimensional array of cameras, an illumination system configured to illuminate a scene with a projected texture, where an image processing pipeline application directs the processor to: utilize the illumination system controller application to control the illumination system to illuminate a scene with a projected texture, capture a set of images of the scene illuminated with the projected texture, and determining depth estimates for pixel locations in an image from a reference viewpoint using at least a subset of the set of images.

Abstract: A three-dimensional (3D) measuring instrument includes a registration camera and a surface measuring system having a projector and an autofocus camera. For the instrument in a first pose, the registration camera captures a first registration image of first registration points. The autofocus camera captures a first surface image of first light projected onto the object by the projector and determines first 3D coordinates of points on the object. For the instrument in a second pose, the registration camera captures a second registration image of second registration points. The autofocus camera adjusts the autofocus mechanism and captures a second surface image of second light projected by the projector. A compensation parameter is determined based at least in part on the first registration image, the second registration image, the first 3D coordinates, the second surface image, and the projected second light.

Abstract: For segmentation in medical imaging, a shape generative adversarial network (shape GAN) is used in training. By including shape information in a lower dimensional space than the pixels or voxels of the image space, the network may be trained with a shape loss or optimization. The adversarial loss and the shape loss are used to train the network, so the resulting generator may segment complex shapes in 2D or 3D. Other optimization may be used, such as using a loss in image space.

Abstract: Methods and systems for measuring 3-dimensional distances on medical images are described. One example method includes receiving at least a first 2-dimensional image on a first plane and a second 2-dimensional image on a second plane, wherein the first plane and the second plane are different planes, selecting a first point on the first 2-dimensional image as a start point, selecting a second point on the second 2-dimensional image as an end point, determining a shortest path between the start point and the end point, performing a measurement of the shortest path between the start point and the end point, and, taking an action based on the measurement.

Abstract: The present disclosure relates to methods and apparatuses for dynamic action detection and dynamic action control. A dynamic action detection method includes: adding an image frame in a video stream into a first queue, to obtain the first queue with a partially updated image frame; detecting a dynamic action in the partially updated image frame; and in response to the dynamic action not matching with the action detection result, updating an action detection result according to the dynamic action, the action detection result including an action reference result or a previously detected action detection result. A dynamic action detection in the present disclosure is executed according to the partially updated image frame in the first queue, so that the dynamic action can be detected more timely; and the dynamic action can also be rapidly determined, so that the action detection result is more accurate and the detection efficiency is higher.

Abstract: A method for medical image segmentation. The method includes accessing and updating a knowledge-base in accordance with embodiments of the present invention. The techniques include: receiving a medical image and computing a sparse landmark signature based on the medical image content. Next, a knowledge-base is searched for representative matches to form a base set, wherein the base set comprises a plurality of reference image sets. A portion of the plurality of reference image sets of the base set is deformed to generate mappings from the base set to the medical image set. Finally a weighted average segmentation for each structure of interest of the medical image set is determined.

Abstract: Various embodiments are generally directed to techniques for image content extraction. Some embodiments include extracting contextually structured data from document images, such as by automatically identifying document layout, document data, document metadata, and/or correlations therebetween in a document image, for instance. Several embodiments include extracting contextually structured data from table images, such as gridded and non-gridded tables. For example, the contents of cells may be extracted from a table image along with structural context including the corresponding row and column information. Many embodiments are directed to generating and utilizing a document template database for automatically extracting document image contents into a contextually structured format. Several embodiments are directed to automatically identifying and associating document metadata with corresponding document data in a document image to generate a machine-facilitated annotation of the document image.

Abstract: According to one embodiment, an image analysis device includes one or more processors configured to receive input of an image; calculate feature amount information indicating a feature of a region of the image; recognize a known object from the image on the basis of the feature amount information, the known object being registered in learning data of image recognition; recognize a generalization object from the image on the basis of the feature amount information, the generalization object being generalizable from the known object; and output output information on an object identified from the image as the known object or the generalization object.

Abstract: Various embodiments for enforcing safe operation of machinery performing an activity in a three-dimensional (3D) workspace includes computationally generating a 3D spatial representation of the workspace; computationally mapping 3D regions of the workspace corresponding to space occupied by the machinery and a human; and based thereon, restricting operation of the machinery in accordance with a safety protocol during physical performance of the activity.

Abstract: The present invention proposes a method for producing an authenticable moiré shape that simultaneously moves and shows a beating effect. The method relies on a combination of the 1D or the 2D moiré and the level line moiré. When tilting a compound showing such a moiré, the moiré shape moves, its intensity levels change significantly but its shape remains the same and is recognizable. Embodiments comprise a base layer made of patterned metallic tiny shapes and a revealing layer made of a 1D array of cylindrical lenslets or of a 2D array of spherical or aspherical lenslets.

Abstract: A robot control device is configured to control a robot operating in a shared work area with an operator. The robot control device includes a distance estimation unit configured to obtain a reachable range of the operator on the basis of information on position and information on physical characteristics of the operator, and obtain a distance between the reachable range and the robot on the basis of information on position and information on shape of the robot and the reachable range, an interference probability calculation unit configured to obtain an interference probability between the robot and the operator on the basis of the distance, and an operation control unit configured to control operation of the robot on the basis of the interference probability.

Abstract: A method of determining facial metrics of a patient includes: capturing a first image of the patient with an image capturing device positioned in a first known position with the assistance of a positioning device, capturing a second image of the patient with the image capturing device positioned in a second known position different that the first known position with the assistance of the positioning device, and determining facial metrics of the patient by analyzing the first image and the second image.

Abstract: The present invention discloses a visual relationship detection method based on a region-aware learning mechanism, comprising: acquiring a triplet graph structure and combining features after its aggregation with neighboring nodes, using the features as nodes in a second graph structure, and connecting in accordance with equiprobable edges to form the second graph structure; combining node features of the second graph structure with features of corresponding entity object nodes in the triplet, using the combined features as a visual attention mechanism and merging internal region visual features extracted by two entity objects, and using the merged region visual features as visual features to be used in the next message propagation by corresponding entity object nodes in the triplet; and after a certain number of times of message propagations, combining the output triplet node features and the node features of the second graph structure to infer predicates between object sets.