Abstract: An image segmentation system is discloses that provides one or more possible contours of a feature of an image, in parallel, that respectively correspond to different interpretations of the image. First, a relatively large set of possible contours are generated in accordance with an image segmentation algorithm. Subsequently, this set of possible contours is reduced to a few candidates reflecting representative solutions corresponding to one or more desired applications.

Abstract: A detection apparatus includes a memory and a processor coupled to the memory. The processor configured to determine a position of each of a plurality of partial images relative to the wide-angle image, to extract, from among the plurality of partial images, a pair of partial images that are consecutive in an image-capturing order, that do not have an overlapping portion, and at least one of which includes an image of a damaged portion, to detect a region of the outside of the partial image to which a damaged portion is estimated to be continuous, as an image-capturing omission candidate region, to determine the image-capturing omission candidate region as an image-capturing omission region, in a case where the image-capturing omission candidate region is included in none of the other partial images, and to issue, based on the position, a notification of a position corresponding to the image-capturing omission region.

Abstract: A method is disclosed. The method comprises defining a region of interest on a three-dimensional model of an object. The method may comprise capturing an image of an object manufactured according to the three-dimensional model. The method may comprise determining, using a processor, a position and orientation of the three-dimensional model with respect to the object in the captured image. The method may comprise locating the region of interest on the manufactured object. A portion of the manufactured object within the region of interest may comprise an identifiable feature. An apparatus and a machine-readable medium are also disclosed.

Abstract: In a verification method for verifying a verification target camera, an original image in which optical characteristics of an original image camera different from the verification target camera are corrected to a captured image of a real scene captured by the original image camera is prepared, a verification image in which an influence of optical characteristics of the verification target camera is applied to the original image is prepared, image recognition of the verification image is performed by executing a verification target algorithm applied for image recognition of an image captured by the verification target camera, and an image recognition result of the verification image is evaluated.

Abstract: A system and method for refocusing an image including determining a point spread function (PSF) according to region of interest (ROI) and pixel depth, and converting the PSF to a gradient domain including differentiating the PSF to reduce nonzero elements in the PSF. The technique spreads intensity of pixels into a circle of confusion per the differentiated PSF. A shape of an optical system or aperture of the imaging device may be considered.

Abstract: According to an embodiment, an information processing apparatus includes an attribute determiner and a setter. Each of acquired first sets indicates a combination of observation information indicating a result of observation of an area surrounding a moving body and position information. The attribute determiner is configured to determine, based on the observation information, an attribute of each of areas into which the area surrounding the moving body is divided, and to generate second sets, each indicating a combination of attribute information indicating the attribute of each area and the position information. The setter is configured to set, based on the second sets, reliability of the attribute of the area of a target second set, from correlation between the attribute of the area of the target second set and the attribute of corresponding areas each indicating the area corresponding to the target area in the areas of the other second sets.

Abstract: The application relates to a video recommendation method and device, and a computer readable storage medium. The video recommendation method comprises: obtaining a user feature of a sample user and a video feature of a sample video; learning a click rate, a like rate, and a follow rate on the basis of a full-connection neural network algorithm to obtain trained user feature and video feature; performing, according to the trained user feature and video feature, combined learning on the click rate, the like rate, and the follow rate on a user side neural network and a video side neural network; and obtaining a video recommendation list according to a network parameter of a neural network algorithm obtained by means of combined learning. According to the video recommendation method, by adding a full-connection neural network algorithm training phase, the trained user feature and video feature are obtained.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating, based on a portrait image, a foreground image mask to indicate foreground pixels of the portrait image; identifying a percentage of white or near white pixels in the foreground by using the foreground image mask and pixel colors in the portrait image; determining whether the percentage of white or near white pixels in the foreground is larger than a predefined threshold; in response to determining, triggering identification of edge pixels in a background of the portrait image; adjusting white background pixels to add shadows by darkening the white background pixels; and adjusting the white or near white pixels in the foreground by darkening the white or near white pixels.

Abstract: An image processing device that detects an image of an object from input data captured by a vision sensor, on the basis of a model pattern including a set of a plurality of first feature points representing the shape of the object, includes an object detection unit that detects images of the object by matching between a plurality of second feature points extracted from the input data and a plurality of first feature points forming the model pattern, a corresponding point selection unit that selects, for each image of the object, second feature points corresponding to the first feature points and stores the selected second feature points as corresponding points, and a model pattern correction unit that calculates a statistic of a predetermined physical quantity of the plurality of corresponding points associated with each first feature point and corrects the first feature point on the basis of the statistic.

Abstract: An information processing apparatus according to one embodiment includes a processing circuit. The processing circuit calculates a first presence probability of an object present around a moving body with positional information measured by each of a plurality of sensors having different characteristics, acquires non-measurement information indicating that the positional information on the object has not been obtained for each of the sensors, and determines a second presence probability of the object based on the first presence probability and the non-measurement information.

Abstract: A method and system for automatically adapting an image data set obtained by an X-ray device. Binary masks are generated for single images of the image data set. The masks differentiate different components of an imaged target object from each other. A reference image of the image data set is determined therefrom, that contains a largest cohesive region of one of the components. From a mean value determined for the region and a specified target value, an offset is determined in such a way that by applying it to the pixel values used for determining the mean value. The pixel values are adapted in such a way that a mean value of the pixel values adapted by the offset corresponds to the specified target value. The determined offset is then applied to all pixel values of the image data set in order to adapt the image data set.

Abstract: In an illustrative embodiment, systems and methods for automating candidate video assessments include receiving a submission from a candidate for an available position including baseline response video segments and question response video segments. The system can determine, from detected nonverbal features within the baseline response video segments, nonverbal baseline scores. For each of the interview questions, candidate response attributes can be detected including a response direction, a response speed, and nonverbal features. A nonverbal reaction score is calculated from the detected nonverbal features and the baseline scores. A response score can be calculated from the response direction and response speed, and a trustworthiness score is determined based on a correspondence between the response score and the nonverbal reaction score. A next interview question can be determined in real-time from a benchmarked version of the response score.

Abstract: The present invention generally relates to a method for transitioning a device controller comprised with an electronic device from an at least partly inactive mode to an at least partly active mode, the electronic device further comprising a pre-processing module and a fingerprint sensor configured to acquire image data. The invention also relates to a corresponding electronic device and to a computer program product.

Abstract: An object appearance detection system with posture detection and a control method thereof are provided. A controlling and computing device uses a first sensor and a second sensor to control a conveying production line, and controls a robotic arm to move an object to be detected to a posture detection position and a surface detection position. When the object to be detected is in the posture detection position, the controlling and computing device receives a posture detection image to perform posture detection on the object to be detected. When the object to be detected is in the surface detection position, the controlling and computing device controls the robotic arm to adjust the posture of the object to be detected according to the posture detection result and receives images from a remote imaging device and photographing devices to perform surface defect detection on the object to be detected.

Abstract: According to an embodiment, an information processing apparatus includes a memory having computer executable components stored therein; and a processing circuit communicatively coupled to the memory. The processing circuit acquires a plurality of pieces of observation information of surroundings of a moving body, generates a plurality of pieces of attribute information of the surroundings of the moving body on the basis of the plurality of pieces of observation information, and sets a reliability of the attribute information of the surroundings of the moving body on the basis of correlation of the plurality of pieces of attribute information.

Abstract: Captured samples of a physical structure or other scene are mapped to a predetermined multi-dimensional coordinate space, and spatially-adjacent samples are organized into array cells representing subspaces thereof. Each cell is classified according to predetermined target-identifying criteria for the samples of the cell. A cluster of spatially-contiguous cells of common classification, peripherally bounded by cells of different classification, is constructed, and a boundary demarcation is defined from the peripheral contour of the cluster. The boundary demarcation is overlaid upon a visual display of the physical scene, thereby visually demarcating the boundaries of a detected target of interest.

Abstract: A placement detection system includes: a placement table; an imaging device that images an object placed on the placement table to generate an input image; and a control device. The control device generates a first binarized image for the input image based on a first threshold value and determines whether the object is placed in a predetermined placement orientation. The control device changes a threshold value for a target pixel to a second threshold value higher than the first threshold value when the object is determined to be placed in the predetermined placement orientation, the target pixel being sandwiched between pixels having pixel levels lower than or equal to the first threshold value and has a higher pixel level than the first threshold value in a first region, generates a second binarized image for the target pixel based on the second threshold value, and detects a shape of the object.

Abstract: A method in an imaging controller of detecting depth sensor artifacts includes: obtaining, from first and second sensors, first and second pluralities of points defined by respective (i) planar positions and depths in a common frame of reference, and (ii) scan angles relative to field of view centers of the first or second sensors; for each of a subset of candidate points from the first plurality of points: searching the second plurality of points for a validator point having (i) a planar position within a threshold distance of a planar position of the candidate point, and (ii) a scan angle smaller than a scan angle of the candidate point; responsive to identifying the validator point: when the depth of the validator point exceeds the depth of the candidate point, classifying the candidate point as an artifact.

Abstract: Systems, methods, and articles of manufacture are disclosed that enable the compression of depth data and real-time reconstruction of high-quality light fields. In one aspect, spatial compression and decompression of depth images is divided into the following stages: generating a quadtree data structure for each depth image captured by a light field probe and difference mask associated with the depth image, with each node of the quadtree approximating a corresponding portion of the depth image data using an approximating function; generating, from the quadtree for each depth image, a runtime packed form that is more lightweight and has a desired maximum error; and assembling multiple such runtime packed forms into per-probe stream(s); and decoding at runtime the assembled per-probe stream(s). Further, a block compression format is disclosed for approximating depth data by augmenting the block compression format 3DC+(BC4) with a line and two pairs of endpoints.

Abstract: In accordance with some embodiments, a training method for biometric identity and authentication is provided. The method includes obtaining biometric data from a plurality of sources. The method further includes establishing a candidate set of neural network parameters. The method additional includes extracting a plurality of feature vectors from the biometric data using the candidate set of neural network parameters. The method also includes determining whether or not the plurality of feature vectors match a training vector set within an error threshold. The method further includes updating the candidate set of neural network parameters in response to determining that the plurality of feature vectors do not match the training vector set within the error threshold.