1949 results about "Recognition algorithm" patented technology

A gaze tracking technique is implemented with a head mounted gaze tracking device that communicates with a server. The server receives scene images from the head mounted gaze tracking device which captures external scenes viewed by a user wearing the head mounted device. The server also receives gaze direction information from the head mounted gaze tracking device. The gaze direction information indicates where in the external scenes the user was gazing when viewing the external scenes. An image recognition algorithm is executed on the scene images to identify items within the external scenes viewed by the user. A gazing log tracking the identified items viewed by the user is generated.

Systems and methods for two-way, interactive communication regarding emergency notifications and responses for mobile environments are disclosed. In accordance with one embodiment of the present invention, a specific geographic area is designated for selective emergency communications. The emergency communications may comprise text, audio, video, and other types of data. The emergency notification is sent to users' mobile communications devices such as in-vehicle telematics units, cellular phones, personal digital assistants (PDAs), laptops, etc. that are currently located in the designated area. The sender of the emergency message or the users' service provider(s) may remotely control cameras and microphones associated with the users' mobile communications devices. For example, a rear camera ordinarily used when driving in reverse may be used to capture images and video that may assist authorities in searching for a suspect. The users' vehicles may send photographs or video streams of nearby individuals, cars and license plates, along with real-time location information, in response to the emergency notification. Image recognition algorithms may be used to analyze license plates, vehicles, and faces captured by the users' cameras and determine whether they match a suspect's description. Advantageously, the present invention utilizes dormant resources in a highly beneficial and time-saving manner that increases public safety and national security.

A phishing detection agent is provided. In one embodiment, a user's browser includes a plug-in application or agent that may capture a visual record of a webpage and, with a cached copy of known, authentic websites provided to it via periodic updates, perform a series of image comparison functions to determine if the suspected website is attempting to deceive the user. The phishing detection agent is capable of performing an image recognition algorithm, such as logo recognition algorithm, optical character recognition, an image similarity algorithm, or combination of two or more of the above. If the suspected webpage corresponds to one of the authentic web pages, but the domain name of the suspected web page does not match the domain name of one of the authentic web pages, the suspected web page is flagged as a phishing web site.

The personalized content system of the system combines a summary music identification value creation and identification algorithm that represents a mathematical summary music identification of a song, an audio file, or other relational music criteria and data (e.g. title, artist, genre, style, beats per minute, etc.) or any combination thereof. The derived value represents the musical taste or style attributes of a song or style. The analysis and generation of the summary music identification value is intended to take place while outside the vehicle where attention can be paid and vehicle distraction and safety is not an issue. A user can generate a single summary music identification value or can generate multiple summary music identification values for different criteria. In other cases, summary music identification values can be defined for genre, but with a much more personal touch possible than with fixed playlist radio stations. In another embodiment, a user can obtain and use someone else's summary music identification value (e.g. a celebrity, artist, friend, etc.) and use it with the system.

Systems and methods for recognizing the gestures of an entity, notably a human being, and, optionally, for controlling an electrical or electronic system or apparatus, are discussed. The system uses sensors that measure signals, preferentially representative of inertial data about the movements of said entity, and implements a process for enriching a dictionary of said gestures to be recognized and a recognition algorithm, for recognition among the classes of gestures in said dictionary. The algorithm implemented is preferably of the dynamic time warping type. The system carries out preprocessing operations, such as the elimination of signals captured during periods of inactivity of the entity, subsampling of the signals, and normalization of the measurements by reduction, and preferentially uses, to classify the gestures, specific distance calculation modes and modes for merging or voting between the various measurements by the sensors.

The invention discloses a cross-camera pedestrian detection tracking method based on depth learning, which comprises the steps of: by training a pedestrian detection network, carrying out pedestrian detection on an input monitoring video sequence; initializing tracking targets by a target box obtained by pedestrian detection, extracting shallow layer features and deep layer features of a region corresponding to a candidate box in the pedestrian detection network, and implementing tracking; when the targets disappear, carrying out pedestrian re-identification which comprises the process of: after target disappearance information is obtained, finding images with the highest matching degrees with the disappearing targets from candidate images obtained by the pedestrian detection network and continuously tracking; and when tracking is ended, outputting motion tracks of the pedestrian targets under multiple cameras. The features extracted by the method can overcome influence of illuminationvariations and viewing angle variations; moreover, for both the tracking and pedestrian re-identification parts, the features are extracted from the pedestrian detection network; pedestrian detection, multi-target tracking and pedestrian re-identification are organically fused; and accurate cross-camera pedestrian detection and tracking in a large-range scene are implemented.

Providing enhanced video content includes processing at least one video feed through at least one spatiotemporal pattern recognition algorithm that uses machine learning to develop an understanding of a plurality of events and to determine at least one event type for each of the plurality of events. The event type includes an entry in a relationship library detailing a relationship between two visible features. Extracting and indexing a plurality of video cuts from the video feed is performed based on the at least one event type determined by the understanding that corresponds to an event in the plurality of events detectable in the video cuts. Lastly, automatically and under computer control, an enhanced video content data structure is generated using the extracted plurality of video cuts based on the indexing of the extracted plurality of video cuts.

An input system for mobile search and a method therefor are provided. The input system includes an input module receiving a code input for a specific term and a voice input corresponding thereto, a database including a glossary and an acoustic model, wherein the glossary includes a plurality of terms and a sequence list, and each of the terms has a search weight based on an order of the sequence list, a process module selecting a first number of candidate terms from the glossary according to the code input by using an input algorithm and obtaining a second number of candidate terms by using a speech recognition algorithm to compare the voice input with the first number of candidate terms via the acoustic model, wherein the second number of candidate terms are listed in a particular order based on their respective search weights, and an output module showing the second number of candidate terms in the particular order for selecting the specific term therefrom.

A system and method of audio processing provides enhanced speech recognition. Audio input is received at a plurality of microphones. The multi-channel audio signal from the microphones may be processed by a beamforming network to generate a single-channel enhanced audio signal, on which voice activity is detected. Audio signals from the microphones are additionally processed by an adaptable noise cancellation filter having variable filter coefficients to generate a noise-suppressed audio signal. The variable filter coefficients are updated during periods of voice inactivity. A speech recognition engine may apply a speech recognition algorithm to the noise-suppressed audio signal and generate an appropriate output. The operation of the speech recognition engine and the adaptable noise cancellation filter may advantageously be controlled based on voice activity detected in the single-channel enhanced audio signal from the beamforming network.

A processor integrating and controlling at least two A/V devices by constructing a control model, referred to as a filter graph, of the at least two A/V devices as a function of a physical connection topology of the at least two A/V devices and a desired content to be rendered by one of the at least two A/V devices. The filter graph may be constructed as a function of at least two device filters corresponding to the at least two A/V devices, in which the device filters include certain characteristics of the at least two A/V device. These characteristics may include the input or output pins for each device, the media type that the A/V device may process, the type of functions that the device may serve, etc. The desired content may be received as a user input which is entered via a keyboard, mouse or other comparable input devices. In addition, the user input may be entered as a voice command, which may be parsed by the processor using conventional speech recognition algorithms or natural language processing to extract the necessary information. Once the filter graph is constructed, the processor may control the at least two A/V devices via the filter graph by invoking predetermined operations on the filter graph resulting in the appropriate commands being sent to the at least two A/V devices, thereby results in the rendering of the desired content.

The present invention relates to methods for evaluating the probability that a patient's diagnosis may be treated with a particular clinical regimen or therapy.

Embodiments of the invention perform assisted tagging of images, including tagging of people, locations, and activities depicted in those images. A batch of images is received comprising images of faces, including at least some faces that have not yet been tagged. A facial recognition algorithm is applied to the faces to determine matching data comprising possible tags for each untagged face. A logic engine applies logic rules to reduce the likelihood that certain matches are correct. The most likely match from among the possible matches is selected for suggestion to the user for verification. Once verified, the metadata of the image indicating the recognized people within the image is updated.

The invention discloses a named entities recognition method based on bidirectional LSTM and CRF. The named entities recognition method based on the bidirectional LSTM and CRF is improved and optimizedbased on the traditional named entities recognition algorithm in the prior art. The named entities recognition method based on the bidirectional LSTM and CRF comprises the following steps: (1) preprocessing a text, extracting phrase information and character information of the text; (2) coding the text character information by means of the bidirectional LSTM neural network to convert the text character information into character vectors; (3) using the glove model to code the text phrase information into word vectors; (4) combining the character vectors and the word vectors into a context information vector and putting the context information vector into the bidirectional LSTM neural network; and (5) decoding the output of the bidirectional LSTM with a linear chain condition random field to obtain a text annotation entity. The invention uses a deep neural network to extract text features and decodes the textual features with the condition random field, therefore, the text feature information can be effectively extracted and good effects can be achieved in the entity recognition tasks of different languages.

A system for authentication and/or authorization which incorporates two communication channels, and at least one of third-party data sources, geographic correlation algorithms, speech recognition algorithms, voice biometric comparison algorithms, and mechanisms to convert textual data into speech. A site visitor's identity can be verified using one or all of such features in combination with a visitors address on one of the channels.

A method for searching a collection of digital images on a display screen, comprising: entering an image review mode and displaying on the display screen a first digital image from the collection of digital images; designating a face contained in the first digital image by using an interactive user interface to indicate a region of the displayed first digital image containing the face; using a processor to execute an automatic face recognition algorithm to identify one or more additional digital images from the collection of digital images that contain the designated face; and displaying the identified one or more additional digital images on the display screen.

The present invention discloses a 3D (three-dimensional) convolutional neural network based human body behavior recognition method, which is mainly used for solving the problem of recognition of a specific human body behavior in the fields of computer vision and pattern recognition. The implementation steps of the method are as follows: (1) carrying out video input; (2) carrying out preprocessing to obtain a training sample set and a test sample set; (3) constructing a 3D convolutional neural network; (4) extracting a feature vector; (5) performing classification training; and (6) outputting a test result. According to the 3D convolutional neural network based human body behavior recognition method disclosed by the present invention, human body detection and movement estimation are implemented by using an optical flow method, and a moving object can be detected without knowing any information of a scenario. The method has more significant performance when an input of a network is a multi-dimensional image, and enables an image to be directly used as the input of the network, so that a complex feature extraction and data reconstruction process in a conventional recognition algorithm is avoided, and recognition of a human body behavior is more accurate.

A MMR system for newspaper publishing comprises a plurality of mobile devices, an MMR gateway, an MMR matching unit and an MMR publisher. The MMR matching unit receives an image query from the MMR gateway and sends it to one or more of the recognition units to identify a result including a document, the page and the location on the page. The MMR system also includes a quality predictor as a plug-in installed on the mobile device to filter images before they are included as part of a retrieval request or as part of the MMR matching unit. The quality predictor comprises an input for receiving recognition algorithm information, a vector calculator, a score generator and a scoring module. The quality predictor receives as inputs an image query, context information and device parameters, and generates an outputs a recognizability score. The present invention also includes a method for generating robustness features.

The present invention discloses a parallelized human body behavior identification method. According to the method, skeleton data of Kinect is used as input; a distributed behavior identification algorithm is implemented based on a Spark computing framework; and a complete parallel identifying process is formed. Acquisition of the skeleton data of a human body is based on scene depth acquisition capacity of Kinect and the data is preprocessed to ensure invariability of displacement and scale of characteristics; and a human body structural vector, joint included angle information and skeleton weight bias are respectively selected for static behavior characteristics and a dynamic behavior searching algorithm for a structural similarity is provided. On the identification algorithm, a neural network algorithm is parallelized on Spark; a quasi-newton method L-BFGS is adopted to optimize a network weight updating process; and the training speed is obviously increased. According to an identification platform, a Hadoop distributed file system HDFS is used as a behavior data storage layer; Spark is applied to a universal resource manager YARN; the parallel neural network algorithm is used as an upper application; and the integral system architecture has excellent extendibility.

The invention discloses a mobile terminal capable of entering corresponding scene modes by means of face recognition and an implementation method thereof. The implementation method includes the following steps: scene modes corresponding to people at different age stages are preset on the mobile terminal, and corresponding usable applications and operation interface attributes under the different scene modes are set; when the mobile terminal is unlocked, a camera is utilized to acquire the image information of a current user; face recognition is carried out on the acquired image information to recognize the facial feature of the current user, and according to the recognized facial feature of the current user, the age stage of the current user is analyzed; according to the analyzed age stage of the current user, the mobile terminal is controlled to enter a scene mode corresponding to people at the current age stage. When the invention is adopted, the camera on the mobile phone can be utilized to acquire a face, the age of the current mobile phone user is recognized by a face recognition algorithm, and the mobile phone is controlled to automatically enter a scene mode corresponding to the age of the recognized user.

The invention discloses a picture cropping method and device based on face recognition. The method comprises the steps that face recognition is carried out on a picture; if a human face is recognized, the picture is cropped according to a face recognition result and a target cropping size; if no human face is recognized, body conspicuousness recognition is carried out on the picture, and the picture is cropped according to the body conspicuousness recognition result and the target cropping size. According to the technical scheme, the cropping effect of the picture can be improved, the defect that in the prior art, the picture cropping effect is poor due to the fact that the content of the picture is not recognized or a recognition algorithm is improper is overcome, and the reading experience of a user on a webpage picture is improved.

The present invention discloses a system and method for controlling the response of a device after a whisper, shout, or conversational speech has been detected. In the preferred embodiment, the system of the present invention modifies its speech recognition module to detect a whisper, shout, or conversational speech (which have different characteristics) and switches the recognition algorithm model, and its speech and dialog output. For example upon detection a whisper, the device may change the dialog output to a quieter, whispered voice. When the device detects a shout it may talk back with higher volume. The device may also utilize more visual displays in response to different levels of speech.

A MMR system for newspaper publishing comprises a plurality of mobile devices, an MMR gateway, an MMR matching unit and an MMR publisher. The MMR matching unit receives an image query from the MMR gateway and sends it to one or more of the recognition units to identify a result including a document, the page and the location on the page. The image registration unit includes an indexing unit for generating images adapted to the environment and capabilities of the image capture device. The indexing unit also automatically adapts the configuration of the plurality of recognition units and index tables based upon image queries applied to the plurality of recognition and index tables. The plurality of recognition units and index tables are configured based on content they reference, recognition algorithm used or other factors. The present invention also includes a number of novel methods including a method for generating blurry images for device training; a method for generating classifiers; a method for selecting a pass classifier; and a method for modifying a hypothesizing and testing classifier on unrecognized images.

A system and method of acquiring information from an image of a vehicle in real time wherein at least one imaging device with advanced light metering capabilities is placed aboard a vehicle, a computer processor means is provided to control the imaging device and the advanced light metering capabilities, the advanced light metering capabilities are used to capture an image of at least a portion of the vehicle, and image recognition algorithms are used to identify the current state or position of the corresponding portion of the vehicle.

A system, method and program product for selecting an algorithm and feature set to solve a problem. A perpetual analytics system is disclosed that provides a genetic algorithm for jointly selecting an algorithm and feature set to solve a problem, comprising: an evolutionary computing engine for processing data encoded as chromosomes, wherein each chromosome encodes an algorithm and a feature set; a domain knowledge store that maintains a plurality of algorithms and a plurality of features; a system for applying a generation of chromosomes to a set of data to provide a set of results; and a fitness function for evaluating the set of results to rate a performance of each chromosome in the set of chromosomes; wherein the evolutionary computing engine is adapted to evolve a subset of the set of chromosomes into a new generation of chromosomes.

The invention discloses a lane line predicating and early warning method suitable for a city road. According to the method, firstly, down sampling is performed on a collected road image, a detection region of interest is set and grayed, filtering enhancement is performed on the region of interest, edge detection is performed through a Canny operator, binarization is performed on the image after edge detection, then the image is mapped on a Hough plane, and lane lines are extracted through improved Hough conversion; then, parameters of the lane line are predicated by the adoption of a classical Kalman filtering algorithm, a region of interest is established, and the lane lines in the region of interest are fitted through a least square method to obtain the optimal predication value; finally, road surface information and current position information of automobiles are combined to establish a lane line departure warning model based on transverse distance, and therefore departure warning is achieved. The recognition method for the lane lines is good in real-time performance and robustness, according to the lane line departure warning model, calculation is easy, and warning is accurate.

The invention relates to a method which automatically classifies motion targets learning online, models an image sequence background and detects the motion targets, scene variation, coverage viewing angle and partitioning scene, extracts and clusters characteristic vectors, and marks region classes; the number of the motion targets in a sub-region and certain threshold value initialize Gaussian distribution and prior probability to accomplish initialization of a classifier in accordance with the characteristic vectors of all the motion target regions that pass through the sub-region; the motion targets in the sub-region are classified and parameters of the classifier are online iterated and optimized; classification results in the process of tracking the motion targets are synthesized to output the classification result of the motion result that learns online. The invention is used for detection of abnormalities in monitor scenes, establishing rules for various class targets, enhancing security of monitor system, identifying objects in the monitor scenes, lessening complexity of identification algorithm, improving rate of identification, and for semantized comprehension for the monitor scenes, identifying classes of the motion target and aiding to comprehension for behavior events occurring in the scenes.

The invention details how to capture very high quality face images in demanding environments for either manual review or automated recognition algorithms. By dynamically controlling a host of imaging parameters and segmenting the field of view into regions just around the faces, or one region per face in the field of view, a face image may be generated of superior quality to that which is produced from an imaging system that takes a more global approach to image parameter adjustment. Specific face regions are given imaging priority, at the expense of other regions of lower priority. Furthermore, face regions may be tracked as a function of time and space, thereby sustaining or improving face image quality as the face location migrates with the field of view.

The invention relates to a sensor and unmanned aerial vehicle intelligent perception control technology, so as to realize the intelligent perception of an unmanned aerial vehicle to the surrounding environment. The system can verify other autonomous localization algorithms and target recognition algorithms to improve the research efficiency of the intelligent perception technology. According to the unmanned aerial vehicle intelligent perception system and method based on multiple sensors, the system comprises a laser radar, an RGB-D visual sensor, an IMU inertial measurement unit, an embedded airborne processor and a flight controller; the laser radar is used to measure the distance information of the surrounding environment; the RGB-D visual sensor collects the distance information and image information of the surrounding environment; the laser radar collects 2D point cloud data, while the RGB-D visual sensor collects 3D point cloud data; the IMU inertial measurement unit is a device which measures the three-axis attitude angle and the acceleration of the device; and the embedded airborne processor comprises two independent modules of autonomous positioning and target recognition. The unmanned aerial vehicle intelligent perception system and method based on multiple sensors are mainly used for unmanned aerial vehicle control.

The invention discloses a routing inspection method applied to a high-speed railway machine room. The routing inspection method comprises the following steps that (1), a routing inspection robot is started, and start-up status self-check and equipment initialization are performed; (2), a routing inspection task file is read, the target equipment cabinet number to be detected is obtained, and the position information of equipment cabinets to be detected is exported from the equipment cabinet position database; (3), the routing inspection robot performs real-time positioning according to laser radar data, performs global path planning and local path planning and reaches the target equipment cabinet positions in sequence, when the routing inspection robot reaches the target equipment cabinetpositions, environment detection is performed, and detection is performed on the equipment cabinet state according to a deep learning neural network model and an image recognition algorithm. The invention provides a corresponding routing inspection robot system at the same time, and the routing inspection robot system can be used for performing automatic regular routing inspection on high-speed railway machine room equipment and can finish emergency routing inspection tasks through remote control when an emergency occurs.

The invention discloses an intelligent recognition monitoring system and method for an unmanned substation. Based on a complete IPVS (internet protocol video monitoring system), the monitoring system comprises a front-end video acquisition layer, a video storage layer and a peripheral warning layer, the front-end video acquisition layer performs real-time video monitoring for the substation by the aid of a network camera with an intelligent recognition function and transmits monitoring data to a monitoring center through a network transmission layer, the video storage layer includes substation information on-site scatter storage and monitoring center concentrated storage, the peripheral warning layer acquires environmental information through an infrared/microwave sensor, a vibration detection device and a temperature and humidity acquisition device in real time, and the monitoring center receives abnormal information transmitted by the front-end video acquisition layer, further judges the information, sieves local alarms caused by normal operation and remotely gives an alarm when abnormity really exists. An intelligent recognition module adopts different recognition algorithms for different monitoring objects, the running state of a target device and the running environment of the substation can be rapidly analyzed, and unattended operation of the substation is truly realized.