Abstract: Provided are an image processing apparatus, an image processing method, and an image processing program capable of obtaining an enjoyable image for a user unexpectedly. The user selects images from image groups imaged from January to December 2017 as an image to be printed. From not-selected remaining images, an image having an attribute in which a frequency for the attribute contained in each image is equal to or less than a threshold is found and printed. Since the obtained image is an image irrelevant to the images selected by the user, the image unexpected for the user can be obtained.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for using location data to identify and provide services in association with items appearing in captured images. One of the methods includes receiving, from a device, an image and location data representing the device's physical location, determining, based on the location data, that a particular set of one or more locations are within a threshold distance of the device's physical location, accessing, for each of the one or more locations in the particular set, item information that indicates one or more items that are associated with the location, determining, based on the accessed item information, that the image likely shows a particular item that is associated with one or more locations in the particular set, and providing, to the device, instructions for presentation of information about (i) the particular item and (ii) one or more locations in the particular set that are associated with the particular item.

Abstract: A computer model to identify a type of physical card is trained using simulated card images. The physical card may exist with various subtypes, some of which may not exist or be unavailable when the model is trained. To most robustly identify these subtypes, the training data set for the computer model includes simulated card images that are generated for the card type. The simulated card images are generated based on a semi-randomized background that varies in appearance, onto which an identifying marking of the card type is superimposed, such that the training data for the computer model includes additional randomized sample card images and ensure the model is robust to further variations in subtypes.

Abstract: An image acquisition device, a goods shelf, a monitoring device for a goods shelf, a monitoring method for a goods shelf and an image recognition method are disclosed. The image acquisition device includes at least one reflective part which is configured to form a virtual image of an object through reflection of the object; and a camera which is configured to take a picture of the object and the virtual image of the object, so as to reduce a photograph blind zone.

Abstract: Techniques for generating magnetic resonance (MR) images of a subject from MR data obtained by a magnetic resonance imaging (MRI) system, the techniques including: obtaining input MR data obtained by imaging the subject using the MRI system; generating a plurality of transformed input MR data instances by applying a respective first plurality of transformations to the input MR data; generating a plurality of MR images from the plurality of transformed input MR data instances and the input MR data using a non-linear MR image reconstruction technique; generating an ensembled MR image from the plurality of MR images at least in part by: applying a second plurality of transformations to the plurality of MR images to obtain a plurality of transformed MR images; and combining the plurality of transformed MR images to obtain the ensembled MR image; and outputting the ensembled MR image.

Abstract: Systems and methods are disclosed for evaluating cardiovascular treatment options for a patient. One method includes creating a three-dimensional model representing a portion of the patient's heart based on patient-specific data regarding a geometry of the patient's heart or vasculature; and for a plurality of treatment options for the patient's heart or vasculature, modifying at least one of the three-dimensional model and a reduced order model based on the three-dimensional model. The method also includes determining, for each of the plurality of treatment options, a value of a blood flow characteristic, by solving at least one of the modified three-dimensional model and the modified reduced order model; and identifying one of the plurality of treatment options that solves a function of at least one of: the determined blood flow characteristics of the patient's heart or vasculature, and one or more costs of each of the plurality of treatment options.

Abstract: A method of matching medical images according to user-defined matches rules. In one embodiment, the matched medical images are displayed according user-defined display rules such that the matched medical images may be visually compared in manner that is suitable to the viewer's viewing preferences.

Abstract: An identification device according to one embodiment comprises: an acquisition unit that uses an encoder configured to derive, from data in which a single subject under different conditions has been recorded, feature values as a first feature value derived from data in which a subject to be identified has been recorded; a conversion unit that generates a second feature value by carrying out conversion using the conversion parameter on the first feature value; a discrete classification unit that carries out discrete classification on each of a plurality of third feature values including the second feature value and generates a plurality of discrete classification results indicating the results of the classification; a result derivation unit that derives, on the basis of the plurality of discrete classification results, identification result information; and an output unit that outputs the identification result information.

Abstract: An apparatus for performing 3D estimation on the basis of pictures of at least two different views includes a hypotheses provider, a similarity measure calculator, and a 3D information determiner. The hypotheses provider locally determines 3D information hypotheses for positions of a current picture of a first view on the basis of a pre-estimate which associates a 3D information estimate to each position of a picture of the first view. The similarity measure calculator calculates, for each position of the current picture of the first view, a hypotheses of the respective position by measuring a similarity between a region similarity measure for each of the 3D information of the current picture of the first view at the respective position and a corresponding region of a second view at a position displaced relative to the respective position by a respective 3D information hypothesis.

Abstract: A motion tracking system includes a first image-capturing module, a computing module and a database. The first image-capturing module captures the full body motion of an object to obtain a depth image. The database provides a plurality of training samples, wherein the training samples include a plurality of depth feature information related to joint positions of the object. The computing module receives the depth image, performs an association operation and a prediction operation for the depth image to obtain a plurality of first joint positions of the object. The computing module projects the first joint positions to a three-dimensional space to generate a three-dimensional skeleton of the object. The depth image includes an image in which limbs of the object are not occluded or an image in which some of limbs of the object are not occluded and the others of limbs of the object are occluded.

Abstract: A system for determining an action performed within an input image includes a memory to store one or more instructions, and a processor communicatively coupled to the memory, and configured to execute the one or more instructions in the memory. The processor employs a convolutional neural network (CNN) that includes a predefined number of initial stages for extracting one or more significant features corresponding to the input image, wherein each initial stage includes a first layer, and a residual block, and wherein the first layer is selected from a group consisting of a convolution layer, a max pooling layer, and an average pooling layer. The CNN includes a final stage for classifying the extracted significant features into one or more predefined classes, wherein the final stage is formed of a global average pooling layer, and a dense layer.

Abstract: The present disclosure relates to methods and systems to classify data. A set of classification modules may inspect received data and identify proposed classifications for confidence values for the received data. An aggregation module may receive and aggregate the proposed classifications and confidence values. Based on the aggregated proposed classifications and the confidence values, the aggregation module may generate a final classification for the received data. An external device may perform an action with respect to the received data based on the final classification associated with the data. The action performed may include maintaining the data such that the data may be retrieved upon receipt a request for the data. Any of the classification modules and the aggregation module may be based on training data that may be utilized in subsequent iterations of classifying data to increase classification accuracy.

Abstract: An image processing apparatus includes a two-dimensional image storage unit configured to store a plurality of two-dimensional image data captured by photographing an identical imaging target object under different exposure conditions; a distance image storage unit configured to store distance image data including a pixel array of a known relationship to a pixel array of the two-dimensional image data; a pixel extraction unit configured to extract, among pixels in each of the two-dimensional image data, a first pixel at which a difference in brightness between identical pixels is less than a predetermined value; and a distance image adjusting unit configured to specify a second pixel of the distance image data at a position corresponding to the first pixel in the pixel array, and to set the second pixel as a non-imaging pixel in the distance image data.

Abstract: A method includes obtaining a first image of a scene using a first image sensor of an electronic device and a second image of the scene using a second image sensor of the electronic device. The method also includes generating a first feature map from the first image and a second feature map from the second image. The method further includes generating a third feature map based on the first feature map, the second feature map, and an asymmetric search window. The method additionally includes generating a depth map by restoring spatial resolution to the third feature map.

Abstract: A method for creating navigable views includes receiving digital images, computing a set of feature points for each of the digital images, selecting one of the digital images as a reference image, identifying a salient region of interest in the reference image, identifying other digital images containing a region of interest similar to the salient region of interest in the reference image using the set of feature points computed for each of other digital images, designating a reference location for the salient region of interest in the reference image, aligning the other digital images to the image that contains the designated reference location, ordering the image that contains the designated reference location and the other digital images, and generating a navigable view.

Abstract: An object recognition method and apparatus for a deformed image are provided. The method includes: inputting an image into a preset localization network to obtain a plurality of localization parameters for the image, wherein the preset localization network comprises a preset number of convolutional layers, and wherein the plurality of localization parameters are obtained by regressing image features in a feature map that is generated from a convolution operation on the image; performing a spatial transformation on the image based on the plurality of localization parameters to obtain a corrected image; and inputting the corrected image into a preset recognition network to obtain an object classification result for the image. In the process of the neural network based object recognition, the embodiment of the present application first transforms the deformed image that has deformation, and then performs the object recognition on the transformed image.

Abstract: In a method for performing adaptive content classification of a video content item, frames of a video content item are analyzed at a sampling rate for a type of content, wherein the sampling rate dictates a frequency at which frames of the video content item are analyzed. Responsive to identifying content within at least one frame indicative of the type of content, the sampling rate of the frames is increased. Responsive to not identifying content within at least one frame indicative of the type of content, the sampling rate of the frames is decreased. It is determined whether the video content item includes the type of content based on the analyzing the frames.

Abstract: Some implementations may provide a method that includes: receiving, at a mobile device, data encoding a first digital watermark present on the identification document; extracting a first portion of personally identifiable information encoded by a spatial luminance pattern associated with the first digital watermark; receiving, at a mobile device, data encoding a second digital watermark present on the identification document; extracting a second portion of personally identifiable information encoded by a spatial luminance pattern associated with the first digital watermark; and validating the identification document based on at least one of the first portion of personally identifiable information or the second portion of personally identifiable information.

Abstract: Systems and methods for test object classification are provided in which the test object is docked with a target object in a plurality of different poses to form voxel maps. The maps are vectorized and fed into a convolutional neural network comprising an input layer, a plurality of individually weighted convolutional layers, and an output scorer. The convolutional layers include initial and final layers. Responsive to vectorized input, the input layer feeds values into the initial convolutional layer. Each respective convolutional layer, other than the final convolutional layer, feeds intermediate values as a function of the weights and input values of the respective layer into another of the convolutional layers. The final convolutional layer feeds values into one or more fully connected layers as a function of the final layer weights and input values. The one or more full connected layers feed values into the scorer which scores each input vector to thereby classify the test object.

Abstract: Apparatus, methods and systems of object recognition are disclosed. Embodiments of the inventive subject matter generates map-altered image data according to an object-specific metric map, derives a metric-based descriptor set by executing an image analysis algorithm on the map-altered image data, and retrieves digital content associated with a target object as a function of the metric-based descriptor set.