Abstract: A face detection device for detecting the face of a person in an input image includes the following elements: a face detection circuit including a hardware circuit configured to detect a face in an input image; a signal processing circuit configured to perform signal processing based on an input image signal in accordance with a rewritable program including a face detection program for detecting a face in an input image; and a controller configured to allow the face detection circuit and the signal processing circuit to perform face detection on an image of a frame or on respective images of adjacent frames among consecutive frames, and to control face detection by the signal processing circuit on the basis of a face detection result obtained by the face detection circuit.

Abstract: A method, system and product for locating facial features using symmetry line of the face. The method comprises: obtaining an image of a face of a subject; automatically detecting a symmetry line of the face, wherein the symmetry line intersects at least a mouth region of the face; and automatically locating a facial feature of the face using the symmetry line. Optionally, a rotation of the symmetry line is used to select a template, to rotate a template or to rotate the image. Optionally, the facial feature is symmetrical and the facial feature is searched for using a symmetrical template. Optionally, the automatic locating comprises performing a one dimension correlation of an intensity cross section defined by the symmetry line with a cross section template. Optionally, the automatic location comprises correlating a curve defined by the symmetry line with a template curve.

Abstract: A method, product and system for implementing liveness detector for face verification. A method comprising detecting a symmetry line of the face; and verifying that the subject moved the mouth by computing a score based on values of a pair of images in the symmetry lines, wherein the score is indicative to a difference in the shape of the mouth between the pair of images. Another method comprises: verifying identity of a subject based on facial recognition and voice recognition, said verifying comprise determining there is mouth movement in an image sequence, wherein said determining comprises: in each image of the sequence, detecting a symmetry line of the face; and verifying that the subject moved the mouth, wherein said verifying comprises: computing a score based on comparison of symmetry lines of the face in different images of the set of images; and comparing the score with a threshold.

Abstract: A computer characterizes an area of interest in an event attended by an audience by receiving behavior information for an audience attending the event and event activity information for one or more areas in the event venue. The computer analyzes the audience behavior and event activity to determine the areas in the event venue to which the audience pays attention. The computer determines the objects in the areas in the event venue and determines one or more areas of interest based on the areas in the event venue to which the audience pays attention and the objects in the areas in the event venue to which the audience pays attention. The computer calculates a characteristic of the areas of interest in the event based on: the areas of interest, the objects in the areas in the event venue, and the analyzed audience behavior information related to the areas of interest.

Abstract: A motion analysis apparatus includes an action detection portion that detects a first action performed in correlation with a hit ball direction after a subject hits a ball using measured data which is measured by a sensor unit attached to at least one of golf club and the subject, a hit ball information generation portion that specifies a hit ball direction according to the first action so as to generate hit ball information including the hit ball direction, a motion analysis portion that analyzes motion in which the subject has hit the ball, so as to generate motion analysis information, and a storage processing portion that stores the motion analysis information and the hit ball information in correlation with each other.

Abstract: Systems, methods, and non-transitory computer-readable media can receive a first image including a representation of a first user. A second image including a representation of a second user can be received. A first set of poselets associated with the first user can be detected in the first image. A second set of poselets associated with the second user can be detected in the second image. The first image including the first set of poselets can be inputted into a first instance of a neural network to generate a first multi-dimensional vector. The second image including the second set of poselets can be inputted into a second instance of the neural network to generate a second multi-dimensional vector. A first distance metric between the first multi-dimensional vector and the second multi-dimensional vector can be determined.

Abstract: A wearable device and method are provided for reporting the time based on a wrist-related trigger. In one implementation, a wearable apparatus for providing time information to a user includes a wearable image sensor configured to capture real-time image data from an environment of a user of the wearable apparatus. The wearable apparatus also includes at least one processing device programmed to identify in the image data a wrist-related trigger associated with the user. The processing device is also programmed to provide an output to the user, the output including the time information, based on at least the identification of the wrist-related trigger.

Abstract: This invention provides a system and method for guiding the workpieces to optimal positions to train an assembly system that is generally free of the use of a CMM or similar metrology device. The system and method expresses the image features of the workpieces, when they are in their respective stations, in a common coordinate system. This ability allows a user to visualize the result of assembling the workpieces without actually assembling them, in a “virtual assembly”. The virtual assembly assists guiding placement of workpieces in respective stations into a desired relative alignment. The system and method illustratively generates a composite image using the images from cameras used in guiding the workpieces that helps the user visualize how the part would appear following assembly. The user can reposition the images of workpieces in their respective stations until the composite image has a desired appearance.

Abstract: A system provides automated editing of a media file. Frames of a media file are extracted and feature vectors are generated based thereon. The feature vectors are clustered according to similarity of the content of the feature vectors and the temporal proximity of frames corresponding to the feature vectors in the media file. Clusters are ranked and selected according to the ranking. Segments of the media file corresponding to the selected clusters are combined into a highlight video. Hotspots in a media file may be identified in a media file by detecting local maxima in a measure of movement of pixels between frames. Clusters may be ranked and selected according to an iterative algorithm that identifies clusters that are the most different from an average of the feature vectors and from other selected clusters.

Abstract: In accordance with an embodiment, a method is disclosed for detecting action in a video clip. An audio clip is extracted from the video clip. The audio clip is converted to an auditory spectrogram. The auditory spectrogram is used to construct a self-similarity matrix. The self-similarity matrix is then used to calculate a novelty curve. The clip is then segmented into segments according to peaks in the novelty curve. Each of segments is scored, and then classified as an action clip if the score is above or below a predetermined threshold. Related methods for folding a digitized song into a shorter version of itself and for sequencing a set of user-supplied video and photo clips to a user-supplied song are further disclosed.

Abstract: The invention relates to a method for analysing a multispectral image (10), which includes designing a detection image from the values of a revealing function that quantifies a content shift in the multispectral image between two areas. The revealing function is applied between a target area and a background area, inside a window which is determined around each pixel. The revealing function values are determined from integral images of order one, and optionally also of order two, which in turn are calculated only once initially, so that the total amount of calculations is reduced. The analysis method is compatible with a real-time implementation during a capture of consecutive multispectral images which form a video stream, in particular for an environment-monitoring task.

Abstract: A self-location calculating device projects a patterned light onto a road surface around a vehicle, captures and obtains an image of the road surface around the vehicle. When determining that a condition of the road surface around the vehicle changes as much as or more than a threshold value, the self-location calculating device calculates a current position and a current attitude angle of the vehicle in present by adding an amount of change in the attitude to a current position and a current attitude angle of the vehicle which are calculated in a previous information process cycle.

Abstract: The animal type determination device configured to perform pattern recognition with respect to an image obtained by image-capturing a periphery of a vehicle using a plurality of pre-stored animal image patterns, and to determine the type of an animal in the image is provided. The animal type determination device is configured to recognize the type of a caution target animal indicated by an animal warning sign in the image captured by a camera, raise a determination priority of the animal image pattern corresponding to the type of the caution target animal in a case where the type of the caution target animal is recognized, and performs the pattern recognition of the animal image pattern in the order of the determination priority to determine the type of the animal in the image.

Abstract: A spatiotemporal system and method for parking occupancy detection. The system can generally include suitable image acquisition, processing, transmission and data storage devices configured to carry out the method which includes generating and processing spatiotemporal images to detect the presence of an object in a region of interest, such as a vehicle in a parking stall.

Abstract: The present invention discloses a night parking-monitoring device. It monitors the state change of a parking space at night by tracking at least a light of a vehicle.

Abstract: A system for detecting and classifying information for a motor vehicle is provided. The system includes a sensor mounted within the motor vehicle and a controller in communication with the sensor and having a memory for storing control logic and a processor configured to execute the control logic. The control logic captures optical information from the sensor, classifies the optical information, compares the classified optical information with communication data received by the motor vehicle, generates fused information based on the comparison, and transmits the fused information from the motor vehicle as a source of additional communication data.

Abstract: A surrounding environment recognition device includes an image capturing unit that captures a peripheral image, and a traffic signal recognizing unit that recognizes a traffic signal from within the peripheral image. The image capturing unit captures a plurality of images of frames. The traffic signal recognizing unit recognizes the traffic signal based on a combination of the plurality of images of frames.

Abstract: A surrounding environment recognition device includes an image capturing unit that captures a peripheral image, and a traffic signal detecting unit, which sets a search region with respect to the peripheral image, and detects a traffic signal inside of the search region. The search region becomes smaller as the distance from the image capturing unit is farther away.

Abstract: A method for predicting a line of vision of an occupant of a vehicle. The method includes a step of reading in an initial line of vision of the vehicle occupant, a step of ascertaining a maximum speed of a present head and/or eye movement of the vehicle occupant, starting from the initial line of vision, using sensor data concerning a change in the visual angle of the vehicle occupant representing the present head and/or eye movement, a step of determining an expected magnitude of the present head and/or eye movement, represented with the aid of a predefined model, using the ascertained maximum speed, and a step of using the initial line of vision and the expected magnitude of the present head and/or eye movement to predict the line of vision of the vehicle occupant.

Abstract: A server includes a database that stores template pairs for a plurality of persons, the template pairs each pairing a first template and a second template that are acquired by using first and second polarization methods different from each other, respectively, from a same person, and a collating unit that collates a third template by the first polarization method, the third template being transmitted from a terminal device, with the template pairs for the persons, collates a fourth template by the second polarization method, the fourth template being transmitted from another server, with the template pairs for the persons, and performs personal authentication according to whether a first template pair and a second template pair are an identical template pair, the first template pair including the first template that coincides with the third template, and the second template pair including the second template that coincides with the fourth template.

Abstract: An image processing method for a picture of a participant, photographed in an event, such as a marathon race, increases the accuracy of recognition of a race bib number by performing image processing on a detected race bib area, and associates the recognized race bib number with a person included in the picture. This image processing method detects a person from an input image, estimates an area in which a race bib exists based on a face position of the detected person, detects an area including a race bib number from the estimated area, performs image processing on the detected area to thereby perform character recognition of the race bib number from an image subjected to image processing, and associates the result of character recognition with the input image.

Abstract: The present disclosure relates generally to data hiding for retail product packaging and other printed objects. One embodiment embeds an information signal in a spot color for product packaging. The spot color is screened, and overprinted with process color tint. The tint is modulated prior to overprinting with optimized signal tweaks. The optimization can include consideration of a detector spectral dependency (e.g., red and/or green illumination). Many other embodiments and combinations are described in the subject patent document.

Abstract: A device for decoding magnetic patterns printed on documents comprising a reading head (12) having: a reader (20) arranged to read first magnetic signals belonging to the magnetic patterns and to electromagnetic noise due to sources internal and/or external to the device. The device further comprises: a further reader (40), arranged to read second magnetic signals belonging to the electromagnetic noise, an adder component (25) arranged to algebraically subtract the amplified second magnetic signals from the amplified first magnetic signals, and a converter (16) arranged to convert the resulting signal into a digital signal representing the read magnetic patterns. A method for decoding magnetic patterns is also disclosed.

Abstract: Techniques for improved binarization and extraction of information from digital image data are disclosed in accordance with various embodiments. The inventive concepts include independently binarizing portions of the image data on the basis of individual features, e.g. per connected component, and using multiple different binarization thresholds to obtain the best possible binarization result for each portion of the image data independently binarized. Determining the quality of each binarization result may be based on attempted recognition and/or extraction of information therefrom. Independently binarized portions may be assembled into a contiguous result. In one embodiment, a method includes: identifying a region of interest within a digital image; generating a plurality of binarized images based on the region of interest using different binarization thresholds; and extracting data from some or all of the plurality of binarized images. Corresponding systems and computer program products are also disclosed.

Abstract: A system and method of retrieving an area of interest of an image stored in a storage device. The method includes selecting the area of interest having user desired features, the stored image being divided into a plurality of polygonal image sections, each polygonal image section having substantially homogenous features, the image being further divided into a plurality of quads such that each quad has boundaries that contains at least a portion of a first polygonal image section and at least a portion of a second polygonal image section. The method further includes determining in which quad in the plurality of quads each pixel of the area of interest is located and determining for each quad, in which portion of the first polygonal image section or the second polygonal image section each pixel is located.

Abstract: An apparatus and a method of detecting an object of the apparatus are provided. The apparatus includes a sensing part configured to photograph an image, a storage configured to store a background image frame, and a controller configured to obtain a first difference image and a second difference image from the photographed image and to determine an existence and a position of an object using the first and the second difference images, wherein the first difference image is an image indicating a difference between a currently photographed image frame and a previously photographed image frame, and the second difference image is an image indicating a difference between the currently photographed image frame and the background image frame stored in the storage.

Abstract: Methods and systems are disclosed for increased speed of processing operations on data in two-dimensional arrays, and for detecting a feature in an image. A method for detecting a feature in an image comprises storing, in a set of data memories within a parallel processing system, first image data representing pixels of a first image. The method further comprises storing, in a memory of a host processor coupled to the parallel processing system, feature kernel data representing a set of weight matrices. A method for increased speed of processing operations on data in two-dimensional arrays comprises storing, in a set of data memories within a parallel processing system, first array data representing elements of a first array. The method further comprises, for each of multiple selected elements within the first array, performing a processing operation on the selected element to produce an output element corresponding to the selected element.

Abstract: A handheld scanner incorporates vision software to allow the handheld scanner to be trained for OCR. The handheld scanner can include a user interface to allow a user to associate an image of a mark with electronic data for the mark. The user interface, along with a range finder, can also be used to influence variables that affect the quality of an image scan, thereby improving the quality of results for the image scan and/or decode process. The handheld scanner can also use a font description file during the decode process. The font description file can be generated using a synthetic image file of a character. The synthetic image file can be created by interpreting a marker font file.

Abstract: A method and apparatus for convolutional neural network (CNN) regression based 2D/3D registration of medical images is disclosed. A parameter space zone is determined based on transformation parameters corresponding to a digitally reconstructed radiograph (DRR) generated from the 3D medical image. Local image residual (LIR) features are calculated from local patches of the DRR and the X-ray image based on a set of 3D points in the 3D medical image extracted for the determined parameter space zone. Updated transformation parameters are calculated based on the LIR features using a hierarchical series of regressors trained for the determined parameter space zone. The hierarchical series of regressors includes a plurality of regressors each of which calculates updates for a respective subset of the transformation parameters.

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: A communication device having a function for supplying electric power to a communication partner may include a non-contact control section configured to balanced-output transmission power, an antenna circuit configured to transmit the transmission power as an electromagnetic wave, and a matching circuit which is disposed between an output of the non-contact control section and an input of the antenna circuit and is configured to perform impedance matching. The matching circuit and the antenna circuit are respectively provided with two systems and circuit structures from the output of the non-contact control section to the antenna circuits via the matching circuits of the two systems are formed symmetrically.

Abstract: A wireless IC device includes a wireless IC chip arranged to process a radio signal, a power-supply circuit board that is connected to the wireless IC chip and that includes a power supply circuit including at least one coil pattern, and a radiation plate arranged to radiate a transmission signal supplied from the power-supply circuit board and/or receiving a reception signal to supply the reception signal to the power-supply circuit board. The radiation plate includes an opening provided in a portion thereof and a slit connected to the opening. When viewed in plan from the direction of the winding axis of the coil pattern, the opening in the radiation plate overlaps with an inner area of the coil pattern and the area of the inner area is approximately the same as that of opening.

Abstract: A hang tag may be disclosed. The hang tag can include an RFID tag, a hole, and an upper and lower portion separated by a perforation. The hang tag can additionally be manufactured on a web whereby once the tag is removed from the web, substantially no tackiness may remain on the hang tag.

Abstract: The present invention relates to a data carrier, especially a security document, such as an identification card, e.g. a passport or a driving license, a credit card and a bank card, the data carrier comprising authentication data comprising personalized data, and as a security feature, an array of lenticular elements being arranged to provide only from predetermined angles of view an image of at least a piece of authentication data of said authentication data. In or to make the data carrier impossible or at least very difficult to copy and forge, the array of lenticular elements is provided with a complex structure providing a further security feature.

Abstract: A deep learning network is trained to automatically analyze enterprise data. Raw data from one or more global data sources is received, and a specific training dataset that includes data exemplary of the enterprise data is also received. The raw data from the global data sources is used to pre-train the deep learning network to predict the results of a specific enterprise outcome scenario. The specific training dataset is then used to further train the deep learning network to predict the results of a specific enterprise outcome scenario. Alternately, the raw data from the global data sources may be automatically mined to identify semantic relationships there-within, and the identified semantic relationships may be used to pre-train the deep learning network to predict the results of a specific enterprise outcome scenario.

Abstract: A method of transfer learning includes receiving second data and generating, via a first network, second labels for the second data. In one configuration, the first network has been previously trained on first labels for first data. Additionally, the second labels are generated for training a second network.

Abstract: Systems and methods for training a neural network or an ensemble of neural networks are described. A hyper-parameter that controls the variance of the ensemble predictors is used to address overfitting. For larger values of the hyper-parameter, the predictions from the ensemble have more variance, so there is less overfitting. This technique can be applied to ensemble learning with various cost functions, structures and parameter sharing. A cost function is provided and a set of techniques for learning are described.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training an actor neural network used to select actions to be performed by an agent interacting with an environment. One of the methods includes obtaining a minibatch of experience tuples; and updating current values of the parameters of the actor neural network, comprising: for each experience tuple in the minibatch: processing the training observation and the training action in the experience tuple using a critic neural network to determine a neural network output for the experience tuple, and determining a target neural network output for the experience tuple; updating current values of the parameters of the critic neural network using errors between the target neural network outputs and the neural network outputs; and updating the current values of the parameters of the actor neural network using the critic neural network.

Abstract: A configurable spiking neural network based accelerator system is provided. The accelerator system may be executed on an expansion card which may be a printed circuit board. The system includes one or more application specific integrated circuits comprising at least one spiking neural processing unit and a programmable logic device mounted on the printed circuit board. The spiking neural processing unit includes digital neuron circuits and digital, dynamic synaptic circuits. The programmable logic device is compatible with a local system bus. The spiking neural processing units contain digital circuits comprises a Spiking Neural Network that handles all of the neural processing. The Spiking Neural Network requires no software programming, but can be configured to perform a specific task via the Signal Coupling device and software executing on the host computer.

Abstract: A novel unified neural network framework, the dynamic memory network, is disclosed. This unified framework reduces every task in natural language processing to a question answering problem over an input sequence. Inputs and questions are used to create and connect deep memory sequences. Answers are then generated based on dynamically retrieved memories.

Abstract: Examples of the disclosure enable an information processing system to automatically implement a user interaction diagnostic engine. In some examples, the information processing system analyzes documents to identify parameters associated with profile features and performance metrics. Based on the parameters, correlation values between the profile features and one or more performance metrics are determined. Based on the correlation values, at least one profile feature is identified to improve a parameter associated with the performance metrics. A predictive model associated with the profile features and the performance metrics is generated. One or more actionable items configured to modify the at least one profile feature are determined such that, based on the predictive model, a parameter associated with the performance metrics are predicted to improve upon execution of the one or more actionable items.

Abstract: A design application is configured to perform a system-level optimization of a collection of system components. The design application iteratively executes a multi-objective solver to optimize structural and functional relationships between the system components in order to meet global design criteria and generate a system design. The design application initializes the design process by extracting from a knowledge base system templates having taxonomic, structural, or functional attributes relevant to the system design. The design application generates the knowledge base by mining taxonomic, structural, and functional relationships from a corpus of engineering texts.

Abstract: Disclosed is a method and apparatus for responding to an inquiry from a client via a network. The method and apparatus receive the inquiry from a client via a network. Based on the inquiry, question-answer pairs retrieved from the network are analyzed to determine a response to the inquiry. The QA pairs are not predefined. As a result, the QA pairs have to be analyzed in order to determine whether they are responsive to a particular inquiry. Questions of the QA pairs may be repetitive and similar to one another even for very different subjects, and without additional contextual and meta-level information, are not useful in determining whether their corresponding answer responds to an inquiry.

Abstract: Some embodiments are associated with a receipt, at a feature learning platform, of sensor data associated with normal operation of an industrial asset, the sensor data including values for a plurality of sensors over a period of time. The feature learning platform may extract a plurality of features via hierarchically deep learning, which may capture characteristics of normal operation of the industrial asset and provide the learned features to a classification modeling platform. The classification modeling platform may then create classification models utilizing the learned features, and the classification models may be executed to automatically identify a potential anomaly for an operating industrial asset.

Abstract: Methods, systems and computer program products for predicting breakout artists and media content. Web content is scraped to identify web pages that include the name of an artist or media item and clusters the web pages into one or more headlines. The number of headlines for the artist or media item that occurred during a first time period and the number of content consumers who played the artist or media item during the first time period and during a second time period are counted. A breakout value is, in turn, calculated for the artist or media item using the number of headlines over the first time period, the number of content consumers during the first time period and the number of content consumers during the second time period.

Abstract: An example method includes receiving a data set, generating a topological representation using topological data analysis, at least one metric-lens combination, and the data set, the representation including a plurality of nodes, each of the nodes having one or more data points as members, receiving a new data point, determining distances between the new data point and at least some of the one or more data points, locating the new data point in a location relative to one or more of the nodes using the distances, identifying a subset of the data points closest to the location of the new data point, comparing the subset of the data points to at least some information regarding the new data point to identify a regime, and generating a report indicating a model associating factors associated with the subset of the data points with the new data point for predicting future outcomes.

Abstract: Provided are improved systems, methods, and computer programs to facilitate predictive accuracy for strategic decision support using a query specific Bayesian network. An unconstrained domain model is defined by domain concepts and causal relationships between the domain concepts. Each causal relationship includes a value for the weight of causal belief for the causal relationship. In response to a query, the unconstrained domain model is transformed into a query specific Bayesian network for the domain model by identifying one or more cycles in the unconstrained domain model, eliminating the one or more cycles from the unconstrained domain model; identifying a sub-graph of the unconstrained domain model that is relevant to a query and creating one or more conditional probability tables that comprise the query specific Bayesian network.

Abstract: A computer implemented method to optimize system process is disclosed. The received one or more input data is transformed into a predefined format based on transformation rules. The transformed one or more input data is analyzed at an analyzer, based on one or more pre-defined rules associated with a rule engine. A result associated with an inference is generated at the inference generator, from the analyzed one or more input data, based on automation rules. The generated inference may be one of a positive or a negative. When the generated inference is positive, one or more processes associated with the customer service optimization is simulated at the optimization engine, through a graphical processor. The result of the inference generator and the optimization engine are checked against the one or more processes. The result of the inference generator and the optimization engine may be stored at the data repository.

Abstract: The method includes identifying, by one or more computer processors, a grouping of elements in a storage system. The method further includes identifying, by one or more computer processors, a first element from the identified grouping of elements. The method further includes identifying, by one or more computer processors, a root unit of the first element from the identified grouping of elements. The method further includes determining, by one or more computer processors, a past usage history of computer resources for the identified root unit of the first element. The method further includes calculating, by one or more computer processors, a future usage of computer resources for the identified root unit based upon the determined past usage history.

Abstract: Methods, systems and computer program products for using a selected cohort of content consumers to rate a media object is provided by identifying a media object, determining a first value, where the first value is equal to a number of content consumers who belong to a cohort and who have played the media object, determining a second value. The second value is equal to a total number of content consumers who played the media object. A rating is computed using the first value and the second value.