Abstract: Embodiments of the invention provide a method, system and computer program product for video sentiment analysis in video messaging. In an embodiment of the invention, a method for video sentiment analysis in video messaging includes receiving different video contributions to a thread in a social system executing in memory of a computer and sensing from a plurality of the video contributions a contributor sentiment. Thereafter, a sentiment value for the different video contributions is computed and a sentiment value for a selected one of the video contributions is displayed in a user interface to the thread for an end user contributing a new video contribution to the thread.

Abstract: A method for facial recognition encode analysis comprises providing a training set of Gabor encoded arrays of face images from a database; and, for each encode array in the training set, evaluating the Gabor data to determine the accuracy of the fiducial points on which the encode array is based. The method also comprises training an outlier detection algorithm based on the evaluation of the encode arrays to obtain a decision function for a strength of accuracy of fiducial points in the encode arrays; and outputting the decision function for application to an encode array to be tested.

Abstract: A system and/or method configured to determine a sample frame order for analyzing a video. The video may have multiple frames ordered in a sequence from a beginning to an end. A first sample frame order for analyzing the video may be determined. Determining the first sample frame order may include determining an initial frame for a first iteration, and determining secondary frames for a second iteration. Determining the initial frame and the secondary frames may be based on a function of frame position in the sequence of frames. The initial frame may be associated with a first sample position, and the secondary frames may be associated with secondary sample positions in the sample frame order. A first feature of the video may be determined based on an analysis of the frames in the video performed on the frames in the first sample frame order.

Abstract: RGB-D data generated by at least one optical sensor for a field of view is received. Thereafter, the RGB-D data is bifurcated into (i) RGB data and (ii) depth data for the field of view. One or more bounding polygons are defined within the depth data that each characterize a window within the field of view encapsulating an object. The RGB data is then cropped using the bounding polygon(s). Image processing can later be applied to the cropped RGB data to identify at least one object therein. Related apparatus, systems, techniques and articles are also described.

Abstract: In an image analyzing apparatus, a controller acquires target image data representing a target image having a code image. The code image includes N number of code elements arranged in a first direction. N is an integer equal to or greater than 2. Each of the N number of code elements has two first sides parallel to a second direction orthogonal to the first direction. The controller specifies outline pixels configuring respective outlines of the respective N number of code elements from. The controller specifies a target inclination by analyzing the outline pixels. The target inclination indicates an inclination of the N number of code elements in the target image. The controller specifies a distance between two first sides of each of N number of code elements. The controller determines a representation form of the code image using the distances.

Abstract: The method includes: setting up a hierarchy structure, wherein the hierarchy structure includes more than 2 levels, each slice in a lowest level of the levels is a single fingerprint image generated by a fingerprint sensor, a slice in a second level of the levels includes at most M slices in a first level of the levels, the second level is one level higher than the first level, and M is a positive integer greater than 1; obtaining a new fingerprint image, adding the new fingerprint image into the lowest level, arid updating the hierarchy structure; and outputting an enroll fingerprint image according to a slice in a highest level.

Abstract: System and/or method configured to determine a sample frame order for analyzing a video. The video may have multiple frames ordered in a sequence from a beginning to an end. Segments of the video may be determined based on an analysis of the frames of the video. A first sample frame order for analyzing the video may be determined based on multiple iterations performed on individual ones of the segments of the video. Determining the first sample frame order may include determining initial frames for a first iteration, and determining secondary frames for a second iteration based on a function of frame position in the sequence of frames. The initial frames and the secondary frames may be associated with sample positions in the sample frame order. A first feature of the video may be determined based on an analysis of the frames performed in the first sample frame order.

Abstract: An image processing system for identifying content within image data. The image processing system comprises a processor that is arranged to: arrange image data in a Markov random field (MRF); and calculate state upper bound values of pairs of variables in the image data that are associated with an edge. The calculation of the state upper bound values is performed by the processor assigning the maximum state values of all of the states of a first variable of the pair of variables to the states of a second variable of the pair of variables, identifying the first and second variables from the pair of variables based on a number of states within each of the first and second variables, and determining a single state solution for identifying content in the image data based on the calculation of the state upper bound values.

Abstract: The present disclosure relates to a high precision method, model, and apparatus for calibrating the intrinsic camera parameter values of one or more intrinsic parameters of a camera. The intrinsic parameters include focal lengths, skew, optical center, and radial distortion coefficients. The practical implementation of the solution consists of a camera capturing several images of a planar calibration pattern from different viewpoints, and a computer implemented processing pipeline. The solution does not require additional human input and is robust to misdetection of the calibration pattern in some of the images.

Abstract: Provided are a template selection system, as well as a template selection method, and recording medium storing a template selection program, for selecting a template that will not appear incompatible with a target image when the target image is combined with the template. Specifically, a target image is selected and target image data representing the selected target image is transmitted to an image compositing server. An impression evaluation value of the target image is calculated and templates for which a discrepancy with respect to the calculated impression evaluation value is less than a threshold value are selected. The target image is combined with the templates and image data representing the resulting composite images are transmitted to a smartphone. A desired composite image is selected by the user from among the composite images displayed on the smartphone.

Abstract: A method for constructing a dictionary to represent data from a training data set comprising: modeling the data as a linear combination of columns; modeling outliers in the data set via deterministic outlier vectors; formatting the training data set in matrix form for processing; defining an underlying structure in the data set; quantifying a similarity across the data; building a Laplacian matrix; using group-Lasso regularizers to succinctly represent the data; choosing scalar parameters for controlling the number of dictionary columns used to represent the data and the number of elements of the training data set identified as outliers; using BCD and PG methods on the vector-matrix-formatted data set to estimate a dictionary, corresponding expansion coefficients, and the outlier vectors; and using a length of the outlier vectors to identify outliers in the data.

Abstract: A system and method for segmenting a current frame of a video includes receiving a segmentation mask determined from a previous frame of the video, receiving a background model value and a background deviation value for a given pixel of a previous frame of the video, updating the received background deviation value, updating the background model value independently of the updating of the background deviation value and determining a value of a pixel of a new segmentation mask corresponding to the given pixel based on a foreground model value of the current frame, a segmenting background model value of the current frame and a segmentation threshold, the determining of the new segmentation mask defining whether the given pixel is a foreground pixel or a background pixel of the current frame. The segmentation mask is used for detecting objects, classifying objects and detecting events in the scene captured by the camera.

Abstract: A method and system for detecting and extracting accurate and precise structure in documents. A high-resolution image of documents is segmented into a set of tiles. Each tile is processed by a convolutional network and subsequently by a set of recurrent networks for each row and column. A global-lookup process is disclosed that allows “future” information required for accurate assessment by the recurrent neural networks to be considered. Utilization of high-resolution image allows for precise and accurate feature extraction while segmentation into tiles facilitates the tractable processing of the high-resolution image within reasonable computational resource bounds.

Abstract: In various example embodiments, a system and method are provided for automated estimation of a saliency map for an image based on a graph structure comprising nodes corresponding to respective superpixels on the image, the graph structure including boundary-connecting nodes that connects each non-boundary node to one or more boundary regions. Each non-boundary node is in some embodiments connected to all boundary nodes by respective boundary-connecting edges forming part of the graph. Edge weights are calculated to generate a weighted graph. Saliency map estimation comprises bringing respective nodes for similarity to a background query. The edge weights of at least some of the edges are in some embodiments calculated as a function of a geodesic distance or shortest path between the corresponding nodes.

Abstract: A method is provided for detecting static graphical elements in a sequence of video frames that compares a selected frame in the sequence to each of a plurality of previous frames in the sequence to identify a graphical element that can be a logo. For each pair of frames compared, an absolute difference frame is determined by acquiring an absolute difference value between pixel values for corresponding pixels over at least a portion of the frames in the frame pair. A metric associated with each absolute difference frame is generated, which reflects a degree of dissimilarity. At least some of the absolute difference frames weighted in accordance with the metric associated therewith are summed to generate an accumulation difference frame such that pairs of frames that are more dissimilar have a greater weight. A static graphical element is then identified over a region of the accumulation difference frame in which pixel values satisfy specified criteria.

Abstract: A display device is provided. The display device includes a display configured to display a content, an input unit configured to obtain a photographed image of a user, a storage configured to store feature information of each pre-classified age stage and a processor configured to compare a feature part in the photographed image with the feature information of each age stage stored in the storage, and calculate an age matching probability the feature part with each age stage. The processor selects a plurality of age stages among all age stages by comparing the age matching probability with a preset threshold value, controls the display to display a combined content which includes a plurality of recommended contents corresponding to each of the selected plurality of age stages.

Abstract: Disclosed is an Arabic optical character recognition method using Hidden Markov Models and decision trees, comprising: receiving an input image containing Arabic text, removing all diacritics from the input image by detecting a bounding box of each diacritic and comparing coordinates thereof to those of a bounding box of a text body, segmenting the input image into four layers, and conducting feature extraction on the segmented four layers, inputting results of feature extraction into a Hidden Markov Model thereby generating HMM models for representing each Arabic character, conducting iterative training of the HMM models until an overall likelihood criterion is satisfied, and inputting results of iterative training into a decision tree thereby predicting locations and the classes of the diacritics and producing final recognition results. The invention is capable of facilitating simple recognition of Arabic by utilizing writing feature thereof, and meanwhile featuring comparatively high recognition precision.

Abstract: A music playlist may be generated based on images selected by a user on a computing device. For example, by using image classification and machine learning, image descriptors may automatically be associated with an image and the assigned image descriptors mapped to music descriptors associated with songs. Songs associated with the mapped music descriptors are identified and assembled in a playlist sent to the user's device. Additional images may be identified, for example based on the image descriptors of the selected images and the music descriptors, to identify images to display during playback of the playlist.

Abstract: In order to obtain an image processing apparatus whose object is to realize edge detection with utilizing motion vectors calculated during image encoding, included are: a motion vector processing part to calculate, for each macroblock, information concerning a motion between frames, from motion vectors included in an encoded image signal; a motion vector filter to let through motion vectors each having a predetermined piece of information out of the calculated information; a frame counter to output a count end signal when a counted number of frames of the encoded image signal received by the motion vector processing part reaches a predetermined number of frames; a frequency counter to integrate a frequency of the motion vectors let through the motion vector filter over the predetermined number of frames for each macroblock; and an edge determiner to determine, upon reception of the count end signal, whether or not the frequency integrated by the frequency counter for each macroblock exceeds a threshold, and i

Abstract: Systems and methods for classifying defects using hot scans and convolutional neural networks (CNNs) are disclosed. Primary scanning modes are identified by a processor and a hot scan of a wafer is performed. Defects of interest and nuisance data are selected and images of those areas are captured usa7ing one or more secondary scanning modes. Image sets are collected and divided into subsets. CNNs are trained using the image subsets. An ideal secondary scanning mode is determined and a final hot scan is performed. Defects are filtered and classified according to the final hot scan and the ideal secondary scanning mode. Disclosed systems for classifying defects utilize image data acquisition subsystems such as a scanning electron microscope as well as processors and electronic databases.