Abstract: A method of generating, by means of a processing device, a depth map and/or a fully focused image of a scene, from a series of images of the scene taken under a same viewing angle but with a different focusing, the method including the steps of: a) for each position of a pixel in the series of images, searching for an extremum of the pixel value in the series of images; and b1) assigning to said pixel position a depth value which is a function of the rank, in the series of images, of the image where the extremum has been identified at step a) to construct a depth map of the scene, and/or b2) assigning to said pixel position the value of the extremum identified at step a) to construct a fully focused image of the scene.

Abstract: An image group title assigning device includes: a title candidate assigning means for assigning a plurality of title candidates to an image group having a plurality of images; a characteristic amount calculating means for individually calculating characteristic amounts of a plurality of images included in the image group in association with each of the plurality of title candidates; and a title determining means for determining a title representing the image group from among the title candidates based upon a characteristic amount of each of the images calculated for each of the title candidates.

Abstract: Systems and methods for segmenting video. A segmentation application executing on a computing device receives a video including video frames. The segmentation application calculates, using a predictive model trained to evaluate quality of video frames, a first aesthetic score for a first video frame and a second aesthetic score for a second video frame. The segmentation application determines that the first aesthetic score and the second aesthetic score differ by a quality threshold and that a number of frames between the first video frame and the second video frame exceeds a duration threshold. The segmentation application creates a video segment by merging a subset of video frames ranging from the first video frame to an segment-end frame preceding the second video frame.

Abstract: A method for generating an environmental profile is provided. The method for generating environmental profile includes generating an image of an environment by capturing the environment with at least one recording device, detecting a change of an object in the environment based on the image, and generating an environmental profile based on the change of the object.

Abstract: An electronic system includes a circuitry configured to obtain a sequence of frames of an object under different viewing angles at consecutive time instances. For a first time instance, the circuitry generates a point cloud descriptive for an external surface of the object on basis of (i) a point cloud obtained for a second time instance preceding the first time instance and (ii) disparity information concerning a frame captured at the first time instance.

Abstract: Disclosed is a 3D scanner comprising at least one scanning module for acquiring three-dimensional coordinates of a surface of an object and a positioning device on which the object is placeable or fixable, whereby the positioning device is movable relative to the at least one scanning module, and the 3D scanner further comprising at least one accelerometer, which is configured to measure a change of the position of the positioning device relative to the scanning module, and a first processing unit which is connected to the at least one scanning module and to the positioning device for receiving data. Also disclosed is a method for scanning a surface of an object to acquire three dimensional (3D) coordinates of the surface.

Abstract: Systems, methods, and non-transitory computer-readable media can generate a saliency prediction model for identifying salient points of interest that appear during presentation of content items, provide at least one frame of a content item to the saliency prediction model, and obtain information describing at least a first salient point of interest that appears in the at least one frame from the saliency prediction model, wherein the first salient point of interest is predicted to be of interest to one or more users accessing the content item.

Abstract: A method for processing content in an electronic device and an electronic device for doing the same are provided. The method includes acquiring content including at least one character, and performing at least one of classifying the acquired content into at least one of a plurality of categories by analyzing the acquired content or generating vector images including a vector image corresponding to the at least one character based on the acquired content and displaying at least a part of the vector images on a display functionally connected to the electronic device.

Abstract: In an image analyzing apparatus, a controller in a first analyzing process performs: sequentially identifying line pixel groups from a first side in a first direction; and determining whether a first-type pixel not contiguous to a first subject group constituted by at least one first-type pixel contiguous to each other in a second direction is present in a first region surrounding the first subject group, using first relevant information relating to each line pixel group located on the first side. In a second analyzing process, the controller performs: sequentially identifying the line pixel groups from a second side in the first direction; and determining whether the first-type pixel not contiguous to the first subject group is present in a second region surrounding the first subject group, using second relevant information relating to each line pixel group located on a second side.

Abstract: An information handling system and method includes a display screen for displaying a three dimensional image captured via a three dimensional camera and the processor to detect a selection of a first pixel within the three dimensional image that is proximate to an edge of a first object in the three dimensional image and redefine the selected first pixel to snap to a second pixel within the three dimensional image on the edge of the first object, wherein the second pixel has a large disparity within the three dimensional image and the processor to detect a selection of a third pixel within the three dimensional image that is proximate to an edge of a second object in the three dimensional image and redefine the selected third pixel to snap to a fourth pixel within the three dimensional image on the edge of the second object, wherein the fourth pixel has a large disparity within the three dimensional image.

Abstract: A face recognition system is provided that includes a device configured to capture a video sequence formed from a set of unlabeled testing video frames. The system includes a processor configured to pre-train a face recognition engine formed from reference CNNs on a still image domain that includes labeled training still image frames of faces. The processor adapts the face recognition engine to a video domain to form an adapted engine, by applying non-reference CNNs to domains including the still image and video domains and a degraded image domain. The degraded image domain includes labeled synthetically degraded versions of the frames included in the still image domain. The video domain includes random unlabeled training video frames. The processor recognizes, using the adapted engine, identities of persons corresponding to at least one face in the video sequence to obtain a set of identities. A display device displays the set of identities.

Abstract: A system for automated geospatial image analysis comprising a deep learning model that receives orthorectified geospatial images, pre-labeled to demarcate objects of interest. The module presents marked geospatial images and a second set of unmarked, optimized, training geospatial images to a convolutional neural network. This process may be repeated so that an image analysis software module can detect multiple object types or categories. The image analysis software module receives orthorectified geospatial images from one or more geospatial image caches. Using a multi-scale sliding window submodule, image analysis software scans geospatial images, detects objects present and geospatially locates them.

Abstract: The present disclosure provides systems and methods that enable distance metric learning using proxies. A machine-learned distance model can be trained in a proxy space in which a loss function compares an embedding provided for an anchor data point of a training dataset to a positive proxy and one or more negative proxies, where each of the positive proxy and the one or more negative proxies serve as a proxy for two or more data points included in the training dataset. Thus, each proxy can approximate a number of data points, enabling faster convergence. According to another aspect, the proxies of the proxy space can themselves be learned parameters, such that the proxies and the model are trained jointly. Thus, the present disclosure enables faster convergence (e.g., reduced training time). The present disclosure provides example experiments which demonstrate a new state of the art on several popular training datasets.

Abstract: A system for simplified generation of systems for analysis of satellite images to geolocate one or more objects of interest. A plurality of training images labeled for a study object or objects with irrelevant features loaded into a preexisting feature identification subsystem causes automated generation of models for the study object. This model is used to parameterize pre-engineered machine learning elements that are running a preprogrammed machine learning protocol. Training images with the study are used to train object recognition filters. This filter is used to identify the study object in unanalyzed images. The system reports results in a requestor's preferred format.

Abstract: The purpose of the present invention is to accurately detect, from a remote location without contact, a structure's defects such as cracking, separation, and internal cavities by distinguishing therebetween.

Abstract: Various techniques and tools for encoding and decoding (e.g., in a video encoder/decoder) binary information (e.g., skipped macroblock information) are described. In some embodiments, the binary information is arranged in a bit plane, and the bit plane is coded at the picture/frame layer. The encoder and decoder process the binary information and, in some embodiments, switch coding modes. For example, the encoder and decoder use normal, row-skip, column-skip, or differential modes, or other and/or additional modes. In some embodiments, the encoder and decoder define a skipped macroblock as a predicted macroblock whose motion is equal to its causally predicted motion and which has zero residual error. In some embodiments, the encoder and decoder use a raw coding mode to allow for low-latency applications.

Abstract: Features related to systems and methods for generating a machine learning model that is a composite of at least two other models (e.g., crowd-sourced models contributed by users) are described. Each of the contributed models provide output values that may not be to scale. To account for these differences, a normalization factor for a first machine learning model is generated to adjust values produced by the first machine learning model to correspond with results from the second machine learning model. The crowd-sourced models along with the normalization factor are included in the new image model generated in the claims.

Abstract: An image processing apparatus (600) includes a determiner (607) which determines correction data for each of a plurality of positions in an image based on a point spread function (PSF) relating to each of the plurality of positions, and an image restorer (608) which repeats predetermined image processing by using the correction data to perform image restoration processing, the image restorer is configured to repeat the predetermined image processing N times (N is a positive integer) to perform the image restoration processing, the predetermined image processing includes processing of generating an (n+1)-th intermediate image based on an n-th image (1<n?N), and processing of generating an (n+1)-th image based on the (n+1)-th intermediate image, the n-th image, and the correction data, and the correction data are coefficient data for a difference between the (n+1)-th intermediate image and the n-th image.

Abstract: The present disclosure relates to an image processing apparatus and method and decoding apparatus. The image processing apparatus may comprises a receiver for receiving an image and an image processor for dividing the image into a plurality of regions, and performing filtering by iteratively applying at least one filter to each of the plurality of regions in the image, wherein the at least one filter comprises an asymmetric filter that uses an asymmetric filtering window having height and width in different size.

Abstract: A system, according to various embodiments, receives images such as videos and photographs from one or more portable computing devices associated with one or more individuals while the portable computing devices are in a particular position within a particular location at a particular time. The system determines a virtual position within a 3-D representation of the particular location that generally corresponds to the particular position and combines the images with the 3-D representation to generate an enhanced 3-D representation of the particular location. The system may create one or more avatars to help users better understand how people will look and move within the space. Users may then login to an enhanced 3-D representation application to view a recreation of the activity at a particular time at a particular location.