Abstract: A portion of speech is captured when spoken by a near-end user. A near-end user terminal conducts a communication session, over a network, between the near-end user and one or more far-end users, the session including a message sent to the one or more far-end users. A vetting mechanism is provided via a touchscreen user interface of the near-end user terminal, to allow the near-end user to vet an estimated transcription of the portion of speech prior to being sent to the one or more far-end users in the message. According to the vetting mechanism: (i) a first gesture performed by the near-end user through the touchscreen user interface accepts the estimated transcription to be included in a predetermined role in the sent message, while (ii) one or more second gestures performed by the near-end user through the touchscreen user interface each reject the estimated transcription to be sent in the message.

Abstract: A controller for use in a power converter includes a switch controller coupled to a power switch coupled to an energy transfer element. The switch controller is coupled to receive a current sense signal representative of a drain current through the power switch. The switch controller is coupled to generate a drive signal to control switching of the power switch in response to the current sense signal and a modulated current limit signal to control a transfer of energy from an input to an output of the power converter. A current limit generator is coupled to generate a current limit signal. A jitter generator is coupled to generate a jitter signal. An arithmetic operator circuit is coupled to generate the modulated current limit signal in response to the current limit signal and the jitter signal.

Abstract: The present disclosure is directed to systems, methods, and devices that enable the revising of Optical Character Recognition (OCR) data by indexing and displaying potential error locations within the OCR data. The primary method for revising the OCR data includes a terminal device indexing, displaying, receiving editing operations for, and editing the OCR data. The terminal device is configured to revise OCR data and includes an OCR review element, which, in some embodiments, is a software stored on a non-transitory, computer-readable medium, that is executed by a processing unit to cause the terminal device to index, display, receive editing operations for, and edit the OCR data.

Abstract: A method for a cloud-based feedback-driven image training and recognition includes receiving a set of expert annotations of a plurality of training images of a predetermined subject matter, wherein the expert annotations include a clinical diagnosis for each image or region of interest in an image, training one or more classification models from the set of expert annotations, testing the one or more classification models on a plurality of test images that are different from the training images, wherein each classification model yields a clinical diagnosis for each image and a confidence score for that diagnosis, and receiving expert classification result feedback regarding the clinical diagnosis for each image and a confidence score yielded by each classification model.

Abstract: A medical imaging decision support system is provided that can conduct, and help medical professionals conduct multi-factor brain analysis. Data for disparate processing modes (for example, EEG, MRI, etc.) can be input to the system, processed in parallel in a cloud environment, and the results can be rendered in a thin client (for example, browser) for a user's rapid multi-modal evaluation of a brain.

Abstract: A method for detecting texts included in a specific image is disclosed. The method includes steps of: (a) an apparatus detecting or allowing another device to detect one or more text candidates in the specific image by referring to feature values of pixels in the specific image; (b) the apparatus classifying or allowing another device to classify one or more weak texts in the specific image as strong texts by referring to information on at least one text classified as the strong text in another image related to the specific image if more than a certain percentage of the detected text candidates are classified as the weak texts as a result of comparison between at least one threshold value and at least one feature value of at least one pixel selected in a region where the detected text candidates are included or a value converted from the feature value.

Abstract: In the present disclosure, a client feature and a pre-stored template image feature are obtained; the obtained client feature and template image feature are projected according to a preset projection matrix, to generate a projection feature pair, where the projection matrix is formed by training of a first template image feature of a same object and a second template image feature of a different object; and similarity calculation is performed on the projection feature pair according to a preset similarity calculation rule, to generate a similarity result and prompt the similarity result to a client.

Abstract: An image acquisition unit of a machine learning device acquires n learning images assigned with labels to be used for categorization (n is a natural number larger than or equal to 2). A feature vector acquisition unit acquires a feature vector representing a feature from each of the n learning images. A vector conversion unit converts the feature vector for each of the n learning images to a similarity feature vector based on a similarity degree between the learning images. A classification condition learning unit learns a classification condition for categorizing the n learning images, based on the similarity feature vector converted by the vector conversion unit and the label assigned to each of the n learning images. A classification unit categorizes unlabeled testing images in accordance with the classification condition learned by the classification condition learning unit.

Abstract: A system according to the principles of the present disclosure includes a communication module and a control module. The communication module transmits a first wireless signal indicating a request from a user of a vehicle to adjust operation of a powertrain in the vehicle, a display in the vehicle, and/or a noise actuator in the vehicle. The communication module receives a second wireless signal enabling operation of the powertrain, the display, and/or the noise actuator to be adjusted according to the request. The control module controls the powertrain, the display, and/or the noise actuator based on a first set of values for a control parameter before the second wireless signal is received. The control module controls the powertrain, the display, and/or the noise actuator based on a second set of values for the control parameter after the second wireless signal is received.

Abstract: An image processing device is provided with an acquiring unit configured to acquire a face image and, a control unit configured to specify a face direction in the face image acquired by the acquiring unit and add, based on the specified face direction, a picture expressing a contour of a face component to the face image.

Abstract: A revised support vector machine (SVM) classifier is offered to distinguish between true keywords and false positives based on output from a keyword spotting component of a speech recognition system. The SVM operates on a reduced set of feature dimensions, where the feature dimensions are selected based on their ability to distinguish between true keywords and false positives. Further, support vectors pairs are merged to create a reduced set of re-weighted support vectors. These techniques result in an SVM that may be operated using reduced computing resources, thus improving system performance.

Abstract: The present disclosure discloses a multimedia file push method and apparatus, and relates to the field of network technologies. The method includes: obtaining first visual feature information, where the first visual feature information is visual feature information of a to-be-pushed multimedia file; determining a first user identity according to the first visual feature information, where the first user identity is a push target user identity of the to-be-pushed multimedia file; and pushing the to-be-pushed multimedia file according to the first user identity.

Abstract: A system and method of associating, assigning, and authenticating users with and to a personal protective equipment device is provided. A method can include receiving biometric data, processing the received biometric data, and based on the processing, associating a user with a personal protective equipment device.

Abstract: A system and method for correcting colors in an image in order to identify pills. The system includes a surface that includes a background upon which pills are positioned and a border of known color adjacent to the background. The system also includes a digital camera and a processor. The processor is used to receive and process the image, use target objects on the surface to identify areas of the image that correspond to the known characteristics and features of the surface, perform image diagnostics to determine if the image can be used, and correct image color based on comparison of identified areas of the image that correspond with the known characteristics and features of the surface. Various graphical displays or cues can assist a user in obtaining a usable image.

Abstract: Systems, devices and methods for improved tracking with an electronic device are disclosed. The disclosures employ advanced exposure compensation and/or stabilization techniques. The tracking features may therefore be used in an electronic device to improve tracking performance under dramatically changing lighting conditions and/or when exposed to destabilizing influences, such as jitter. Historical data related to the lighting conditions and/or to the movement of a region of interest containing the tracked object are advantageously employed to improve the tracking system under such conditions.

Abstract: Provided are a fire detecting apparatus and a method thereof for detecting a fire, the method includes operations of extracting a feature of at least one object in an input image by using a value of a brightness difference between pixels of the input image or by using an RGB value of the input image; converting the extracted feature of the at least one object into an N dimensional feature; and performing Support Vector Machine (SVM) machine learning on the N dimensional feature of the at least one object.

Abstract: A region setting unit and a specific region extracting unit are included. The region setting unit sets, within an input image that is photographed at a reference time point out a first region estimated as highly probable to be a specific region and a second region estimated as highly probable to be a background region, which is a region other than the specific region. The specific region extracting unit extracts the specific region within the input image based on a first histogram which is a histogram of density values in the first region and a second histogram which is a histogram of density values in the second region. The specific region extracting unit extracts a specific region from the input image that is photographed at a time point different from the reference time point, based on the first histogram and the second histogram.

Abstract: Exemplary methods, apparatuses, and systems infer a context of a user or device. A computer vision parameter is configured according to the inferred context. Performing a computer vision task, in accordance with the configured computer vision parameter. The computer vision task may by at least one of: a visual mapping of an environment of the device, a visual localization of the device or an object within the environment of the device, or a visual tracking of the device within the environment of the device.

Abstract: A multi-class object classifying method and system are disclosed herein, where the multi-class object classifying method includes the following steps: classes, first training images and second training images are received and stored, and first characteristic images and second characteristic images are respectively extracted from the first training images and the second training images; the first training images is used to generate classifiers through a linear mapping classifying method; a classifier and the second characteristic images are used to determine parameter ranges corresponding to the classes and a threshold corresponding to the classifier.

Abstract: When a feature vector is converted to a reduced vector, a converting unit samples N components of interest from the M components of the feature vector, executes the process of calculating one component of the reduced vector from the N components of interest by d times to create the d-dimensional reduced vector and, the converting unit (1) excludes the components within a predetermined distance D in the same row as the previous component of interest sampled at the previous sampling, (2) excludes the components in the same column as the previous component of interest including the component k rows apart and within the distance (D?k) from the component k rows apart, and (3) samples the component of interest of this time from the remaining components after exclusion when sampling the component of interest.

Abstract: In techniques for image foreground detection, a foreground detection module is implemented to generate varying levels of saliency thresholds from a saliency map of an image that includes foreground regions. The saliency thresholds can be generated based on an adaptive thresholding technique applied to the saliency map of the image and/or based on multi-level segmentation of the saliency map. The foreground detection module applies one or more constraints that distinguish the foreground regions in the image, and detects the foreground regions of the image based on the saliency thresholds and the constraints. Additionally, different ones of the constraints can be applied to detect different ones of the foreground regions, as well as to detect multi-level foreground regions based on the saliency thresholds and the constraints.

Abstract: Mobile friendly internet searches are enabled by determining if individual internet sites are mobile friendly. In a described implementation, a mobile-friendliness indication is determined for an internet site. The determined mobile-friendliness indication is stored in association with the internet site in an index that may be used for internet searching. In another described implementation, a data structure comprises an internet index with multiple internet site entries having one or more associated mobile-friendliness indications. The mobile-friendliness indications may be general affirmative indications of mobile friendliness or specific indications of mobile-friendliness that identify a specific user agent type. In yet another described implementation, a search request having search terms is received, and a search is performed based thereon to produce a search results set.

Abstract: Plural pieces of content are sequentially reproduced, without interruption. A terminal measures a processing time from request for content data of a first tune from a server to decoding of first content data in a streaming format. A reproduction end time of content under reproduction is calculated and a time earlier than the reproduction end time by the processing time is caused to be a request time when content data of a second tune is requested. When the time occurs during reproduction of the first tune, the terminal requests the content data of the second tune from the server. The terminal receives and decodes the content data before the reproduction end time. When the reproduction of the content data of the first tune is ended, the terminal immediately reproduces the content data of the second tune.

Abstract: A computerized recipe station for time-lapse image analysis method includes the steps of inputting an image sequence and an initial recipe to a computer storage; performing by a computer program an incremental apply using the image sequence and the initial recipe to generate an incremental output; pausing the incremental apply; using the incremental output to perform an incremental output assurance operation, which may be an intermediate result analysis to generate an analysis output, a recipe update to generate an updated recipe, or a result editing to generate an edited incremental output; and continuing the incremental apply until pausing or completion to generate a processing output. The analysis output generated by the intermediate result analysis may be used to guide the recipe update step or used to guide the result editing step.

Abstract: A method, device, system, and computer program for object recognition of a 3D object of a certain object class using a statistical shape model for recovering 3D shapes from a 2D representation of the 3D object and comparing the recovered 3D shape with known 3D to 2D representations of at least one object of the object class.

Abstract: An image processor is provided, the image processor including a general-purpose classifier that detects a predetermined large classification target; and a dedicated classifier that detects a small classification target which is a subdivision of the large classification target; an image acquisition part that acquires an image photographed by a camera; an image extracting part that extracts a registered image including a user-assigned domain from the photographed image; a dedicated classifier performing part that causes the dedicated classifier to perform detection processing to the registered image extracted; a dedicated classifier selector that selects the dedicated classifier having a highest index indicating superiority or inferiority of a detection result; and a classifier generator that generates a registered image classifier to detect a target included in the registered image by replacing some weak classifiers included in the selected dedicated classifier with weak classifiers included in the general-pur

Abstract: Methods and systems for predicting light probes for outdoor images are disclosed. A light probe database is created to learn a mapping from the outdoor image's features to predicted outdoor light probe illumination parameters. The database includes a plurality of images, image features for each of the plurality of images, and a captured light probe for each of the plurality of images. A light probe illumination model based on a sun model and sky model is fitted to the captured light probes. The light probe for the outdoor image may be predicted based on the database dataset and fitted light probe models.

Abstract: A method and system for detection of native and bypass coronary ostia in a 3D volume, such as a CT volume, is disclosed. Native coronary ostia are detected by detecting a bounding box defining locations of a left native coronary ostium and a right native coronary ostium in the 3D volume using marginal space learning (MSL), and locally refining the locations of the left native coronary ostium and the right native coronary ostium using a trained native coronary ostium detector. Bypass coronary ostia are detected by segmenting an ascending aorta surface mesh in the 3D volume, generating a search region of a plurality of mesh points on the ascending aorta surface mesh based on a distribution of annotated bypass coronary ostia in a plurality of training volumes, and detecting the bypass coronary ostia by searching the plurality of mesh points in the search region.

Abstract: Dynamic features are utilized with CRFs to handle long-distance dependencies of output labels. The dynamic features present a probability distribution involved in explicit distance from/to a special output label that is pre-defined according to each application scenario. Besides the number of units in the segment (from the previous special output label to the current unit), the dynamic features may also include the sum of any basic features of units in the segment. Since the added dynamic features are involved in the distance from the previous specific label, the searching lattice associated with Viterbi searching is expanded to distinguish the nodes with various distances. The dynamic features may be used in a variety of different applications, such as Natural Language Processing, Text-To-Speech and Automatic Speech Recognition. For example, the dynamic features may be used to assist in prosodic break and pause prediction.

Abstract: A data processing method includes setting a learning data set including a plurality of learning data having respective labels to an uppermost node of a decision tree, setting selection probabilities of the learning data, selecting a combination of the learning data from each of a plurality of sets of learning data assigned to nodes of the decision tree in accordance with the selection probabilities, determining a branch rule for dividing each of the sets of the learning data assigned to the nodes of the decision tree into at least two subsets using the selected combination of the learning data, and assigning the subsets divided in accordance with the branch rule to nodes in a lower level.

Abstract: Methods and systems for reducing the required footprint of SNoW-based classifiers via optimization of classifier features. A compression technique involves two training cycles. The first cycle proceeds normally and the classifier weights from this cycle are used to rank the Successive Mean Quantization Transform (SMQT) features using several criteria. The top N (out of 512 features) are then chosen and the training cycle is repeated using only the top N features. It has been found that OCR accuracy is maintained using only 60 out of 512 features leading to an 88% reduction in RAM utilization at runtime. This coupled with a packing of the weights from doubles to single byte integers added a further 8× reduction in RAM footprint or a reduction of 68× over the baseline SNoW method.

Abstract: Techniques are disclosed relating to automatically adjusting images. In one embodiment, an image may be automatically adjusted based on a regression model trained with a database of raw and adjusted images. In one embodiment, an image may be automatically adjusted based on a model trained by both a database of raw and adjusted images and a small set of images adjusted by a different user. In one embodiment, an image may be automatically adjusted based on a model trained by a database of raw and adjusted images and predicted differences between a user's adjustment to a small set of images and a predicted adjustment based on the database of raw and adjusted images.

Abstract: An interactive concept learning image search technique that allows end-users to quickly create their own rules for re-ranking images based on the image characteristics of the images. The image characteristics can include visual characteristics as well as semantic features or characteristics, or may include a combination of both. End-users can then rank or re-rank any current or future image search results according to their rule or rules. End-users provide examples of images each rule should match and examples of images the rule should reject. The technique learns the common image characteristics of the examples, and any current or future image search results can then be ranked or re-ranked according to the learned rules.

Abstract: A procedure can be assisted by a processor system, such as a computer system. A trajectory can be used to identify a selected trajectory or path of an instrument to reach a tumor within a brain of a subject, reach a selected portion of the anatomy (e.g. sub-thalamic nucleus (STN) or spinal cord), or other appropriate target. The planning algorithm can include both inputted data and learned rankings or ratings related to selected trajectories. The planning algorithm can used the learned ratings to rate and later determined trajectories.

Abstract: An information processing apparatus sets a plurality of reference locations of data in information as one reference location pattern and acquires a feature amount obtained from a value of data of one of the plurality of pieces of reference information in one reference location pattern for each of a plurality of reference location patterns and the plurality of pieces of reference information. The apparatus extracts data included in the input information according to each of the plurality of reference location patterns, selects the reference location pattern for classification of the input information from the plurality of reference location patterns based on a value of data included in the extracted input information, and executes classification of the input information by using the feature amount in the selected reference location pattern and data included in the input information at a reference location indicated by the reference location pattern.

Abstract: Automatic roof identification systems and methods are described. Example embodiments include a roof estimation system configured to automatically detect a roof in a target image of a building having a roof. In one embodiment, automatically detecting a roof in a target image includes training one or more artificial intelligence systems to identify likely roof sections of an image. The artificial intelligence systems are trained on historical image data or an operator-specified region of interest within the target image. Then, a likely outline of the roof in the target image can be determined based on the trained artificial intelligence systems. The likely roof outline can be used to generate a roof estimate report. This abstract is provided to comply with rules requiring an abstract, and it is submitted with the intention that it will not be used to interpret or limit the scope or meaning of the claims.

Abstract: In various example embodiments, a system and method for using machine learning to define user controls for image adjustment is provided. In example embodiments, a new image to be adjusted is received. A weight is applied to reference images of a reference dataset based on a comparison of content of the new image to the reference image of the reference dataset. A plurality of basis styles is generated by applying weighted averages of adjustment parameters corresponding to the weighted reference images to the new image. Each of the plurality of basis styles comprises a version of the new image with an adjustment of at least one image control based on the weighted averages of the adjustment parameters of the reference dataset. The plurality of basis styles is provided to a user interface of a display device.

Abstract: There is provided a person attribute estimation system capable of accurately estimating an attribute of a person according to an environment in which a person who is the target of attribute estimation is to be captured. The person attribute estimation system includes a camera, an attribute estimation unit for estimating an attribute of a person shown in the image generated by the camera, by using an estimation model, a pseudo-site image generation unit for generating a pseudo-site image by processing data of a standard image which is a person image according to image capturing environment data indicating an image capturing environment of the attribute estimation target person by the camera, and an estimation model relearning unit for performing learning of the estimation model by using the pseudo-site image.

Abstract: A system for controlling equipment of a vehicle is provided. The system is configured to image a scene external and forward of the controlled vehicle and to generate image data corresponding to acquired images including a controller for receiving and analyzing the image data and for generating a control signal in response to analysis of the image data and in response to a selected mode of operation. The controller analyzes the data in order to detect at least one characteristic in the image data and selects a mode of operation from at least one of the following modes of operation: the controller selects a dark village mode if at least one characteristic reaches a first threshold, a bright village mode if at least one characteristic reaches a second threshold, and a non-village mode if the controller is not operating in either the dark village mode or the bright village mode.

Abstract: A sequence of images depicting an object is captured, e.g., by a camera at a point-of-sale terminal in a retail store. The object is identified, such as by a barcode or watermark that is detected from one or more of the images. Once the object's identity is known, such information is used in training a classifier (e.g., a machine learning system) to recognize the object from others of the captured images, including images that may be degraded by blur, inferior lighting, etc. In another arrangement, such degraded images are processed to identify feature points useful in fingerprint-based identification of the object. Feature points extracted from such degraded imagery aid in fingerprint-based recognition of objects under real life circumstances, as contrasted with feature points extracted from pristine imagery (e.g., digital files containing label artwork for such objects).

Abstract: A learning apparatus in the present invention includes a weak discriminator generation unit that generates a weak discriminator which calculates a discrimination score of an instance of a target based on a feature and a bag label, a weak discrimination unit which calculates the discrimination score based on the generated weak discriminator, an instance probability calculation unit that calculates an instance probability of the target instance based on the calculated the discrimination score, a bag probability calculation unit that calculates a probability that no smaller than two positive instances are included in the bag based on the calculated instance probability and a likelihood calculation unit which calculates likelihood representing plausibility of the bag probability based on the bag label.

Abstract: Described is a system for object detection using classification-based learning. A fusion method is selected, then a video sequence is processed to generate detections for each frame, wherein a detection is a representation of an object candidate. The detections are fused to generate a set of fused detections for each frame. The classification module generates a classification score labeling each fused detection based on a predetermined classification threshold. Otherwise, a token indicating that the classification module has abstained from generating a classification score is generated. The scoring module produces a confidence score for each fused detection based on a set of learned parameters from the learning module and the set of fused detections. The set of fused detections are filtered by the accept-reject module based on one of the classification score or the confidence score. Finally, a set of final detections representing an object is output.

Abstract: A dictionary of atoms for coding data is learned by first selecting samples from a set of samples. Similar atoms in the dictionary are clustered, and if a cluster has multiple atoms, the atoms in that cluster are merged into a single atom. The samples can be acquired online.

Abstract: An improved branch-and-bound process of interval arithmetic subdivision in furtherance of computation of rigorous error bounds on integrated digital scene information for two dimensional display is provided. More particularly, a first aspect of the subject process includes pseudo-randomly subdividing an interval domain comprising a set of interval variables in furtherance of ascertaining a characteristic contribution of the interval variables of said set of interval variables to an image space comprising at least a sub-pixel area. A further aspect, either alone or in combination with the first aspect contemplates pseudo-randomly discarding a select partitioning of interval variables of a set of interval variables of a geometric function from a computed solution of an interval arithmetic branch-and-bound process.

Abstract: Described is a system for object and behavior recognition which utilizes a collection of modules which, when integrated, can automatically recognize, learn, and adapt to simple and complex visual behaviors. An object recognition module utilizes a cooperative swarm algorithm to classify an object in a domain. A graph-based object representation module is configured to use a graphical model to represent a spatial organization of the object within the domain. Additionally, a reasoning and recognition engine module consists of two sub-modules: a knowledge sub-module and a behavior recognition sub-module. The knowledge sub-module utilizes a Bayesian network, while the behavior recognition sub-module consists of layers of adaptive resonance theory clustering networks and a layer of a sustained temporal order recurrent temporal order network. The described invention has applications in video forensics, data mining, and intelligent video archiving.

Abstract: Systems and methods for developing and using adaptive threshold values for different input images for object detection are disclosed. In embodiments, detector response histogram-based systems and methods train models for predicting optimal threshold values for different images. In embodiments, when training the model, an optimal threshold value for an image is defined as the value that maximizes the reduction of false positive image patches while preserving as many true positive image patches as possible. Once trained, the model may be used to set different threshold values for different images by inputting a detector response histogram for the image patches of an image into the model to determine a threshold value for detection.

Abstract: The present invention uses dynamic grouping to divide up training samples to train different classification nodes. At the beginning of the training, all samples are in the same group. A clustering process is applied in the feature space of the selected feature vectors with cluster indexes accumulated. The average of all the accumulated cluster indexes is used as the threshold for splitting the samples into two groups. When the splitting criterion is met, samples are split into two groups based on their similarity in the feature space.

Abstract: Systems and methods are disclosed for generating a probability density to estimate the probability that an event will occur in a region of interest. The methods input spatial event data comprising one or more events occurring in the region of interest along with auxiliary data related to the region of interest. The auxiliary data comprises non-event data having spatial resolution such that the probability density estimate for the region of interest is calculated based on a function of the auxiliary data and the event data. In particular, the auxiliary data is used to generate a penalty functional used in the calculation of the probability density estimate.

Abstract: An output device has a first input portion that inputs data of a target state, a second input portion that inputs data of a current state, a sensing portion that senses a state of a shift portion that shifts from the current state to another state including the target state and generates sensing data, a transmission portion that transmits the sensing data, the data of the target state, and the data of the current state, to a predetermined external device, a reception portion that receives from the external device suitability determination result data acquired by determination of suitability of the shift portion configured to shift from the current state to the target state, based on the sensing data, the data of the target state, and the data of the current state, and an output portion that outputs the suitability determination result data that the reception portion has received.

Abstract: Certain embodiments of the present invention provide methods and systems for determining a hanging protocol for display of clinical images in a study. Certain embodiments provide a machine learning hanging protocol analysis system. The example system includes an image processing module to process image data to provide one or more features. The example system includes a learning engine to receive processed image data and additional data to learn and adapt a hanging protocol for repeated use by applying one or more machine learning algorithms to the processed image data and additional data. The learning engine is to continue to refine an available selection of candidate layouts based on the processed image data and additional data to provide one or more layout choices for selection to form a hanging protocol for display of image and other data.