Knowledge distillation-based compression method for pre-trained language model, and platform

A knowledge distillation-based compression method for a pre-trained language model, and a platform. In the method, a universal feature transfer knowledge distillation strategy is first designed, and in a process of distilling knowledge from a teacher model to a student model, feature maps of each layer of the student model are approximated to features of the teacher model, with emphasis on the feature expression capacity in intermediate layers of the teacher model for small samples, and these features are used to guide the student model; then, the ability of the self-attention distribution of the teacher model to detect semantics and syntax between words is used to construct a knowledge distillation method based on self-attention crossover; and finally, in order to improve the learning quality of early-period training and the generalization ability of late-period training in the learning model, a Bernouli probability distribution-based linear transfer strategy is designed to gradually complete ...

다수의 층들을 사용하는 재귀적 흐름 가스 분배 스택의 제조

... 방법은 복수의 금속 플레이트들과 연관되는 하나 이상의 파라미터들을 수신하는 것을 포함한다. 방법은, 하나 이상의 파라미터들에 기초하여, 복수의 금속 플레이트들과 연관되는 복수의 예측된 변형 값들을 결정하는 것을 더 포함한다. 복수의 예측된 변형 값들 각각은 복수의 금속 플레이트들 중 대응하는 금속 플레이트에 대응한다. 방법은, 본딩된 금속 플레이트 구조물을 생성하도록, 복수의 예측된 변형 값들에 기초하여, 복수의 금속 플레이트들이 확산 본딩되게 하는 것을 더 포함한다.

이미지 기반 브라우저 내비게이션

... 이미지 데이터에 기초하여 브라우저를 내비게이팅하는 시스템은: 클라이언트 디바이스로부터 스캔 요청을 수신하는 동작- 스캔 요청은 이미지 데이터를 포함하는 이미지를 포함함 -; 이미지 데이터에 기초하여 이미지 내에 묘사된 객체를 식별하는 동작; 객체의 분류를 결정하는 동작; 및 분류에 기초하여 클라이언트 디바이스와 연관된 브라우저를 리소스로 내비게이팅하는 동작을 포함하는 동작들을 수행할 수 있다.

Apparatus and Method for Classifying Electronic Components

... 전자부품 분류장치 및 그 방법을 개시한다. 본 실시예에 의하면, 분류대상 부품을 카메라로 촬영하여 획득한 촬영영상을 분석하고, 분석정보를 기반으로 분류대상 부품의 소재가 금속 속성인지 여부를 판단하여 금속 속성인 경우, 분류대상 부품의 촬영영상을 가공하여 자동 광학 검사기에 정확한 부품 코드(Code)를 제공하여 자동으로 분류대상 부품을 분류하는 전자부품 분류장치 및 그 방법을 제공한다.

Animal identification using vision techniques

A vision system for determining an identity of an individual animal in a population of animals with known identity is provided. A first image sensor arrangement is configured to capture three-dimensional or thermographic images of the individual animal. A second image sensor arrangement is configured to capture two-dimensional images of the individual animal. Control circuitry is configured to obtain a first image of the individual animal captured by the first image sensor arrangement and to detect a body part in the first image, and in response to detecting the body part, obtain a second image of the individual animal captured using the second image sensor arrangement, and to identify the individual animal by comparing metrics of the second image with the reference data in a data storage.

Contextual embeddings for improving static analyzer output

In a computer-implemented method for improving a static analyzer output, a processor receives a labeled data set with labeled true vulnerabilities and labeled false vulnerabilities. A processor receives pretrained contextual embeddings from a contextual embeddings model. A processor maps the true vulnerabilities and the false vulnerabilities to the pretrained contextual embeddings model. A processor generates a fine-tuned model with classifications for true vulnerabilities.

SYSTEM AND METHOD FOR INTERFACING WITH A DECISIONING SERVICE FROM A THIRD PARTY DOMAIN

A system for interfacing with a decisioning service from a third-party domain. A backend system may receive a request to interface with the decisioning service to generate a prequalification result for a user for a specified product, in response to the actuation of a button included in a third-party web domain rendering a product. The backend system may execute a call on a website of the decisioning service. The backend system may receive the product information from the button. The backend system may generate the website within the third-party domain, including a request for information about the user, in response to executing the call to the website. The backend system may transmit a request for generating prequalification result for the product to the decisioning service. The backend system may receive the prequalification result from the decisioning service.

SYSTEM AND METHOD FOR PROVIDING API VERSION CONTROL

Described herein is a system for providing version control across APIs. In an embodiment, an application of a client device may transmit communications to a service application API that are in a format not accepted by the service application API, using the application API. A central (or public) API may receive the communication transmitted by the application's API. The central API may identify the versions of the application's API and the service application API. The central API may format the communication based on the mapped fields, to a format accepted by the service application API. The central API 106 may forward the formatted communication to the service application.

Intelligent sorting of time series data for improved contextual messaging

Systems for intelligent sorting of time series data for improved contextual messaging are included herein. An intelligent sorting server may receive time series data comprising a plurality of chat messages. The intelligent sorting server may determine a first order of the plurality of chat messages based on a chronologic order. The intelligent sorting server may use one or more machine learning classifiers to identify candidates for reordering the chat messages. The intelligent sorting server may generate a second order of the chat messages based on the identified candidates for reordering. Accordingly, the intelligent sorting server may present, to a client device, a transcript of the chat messages associated with the second order and an indication that at least one chat message has been repositioned.

SYSTEM AND METHOD FOR DETERMINING TARGET FEATURE FOCUS IN IMAGE-BASED OVERLAY METROLOGY

A metrology system includes one or more through-focus imaging metrology sub-systems communicatively coupled to a controller, where the controller includes one or more processors. The one or more processors may be configured receive a plurality of training images captured at one or more focal positions. The one or more processors may further generate a machine learning classifier based on the plurality of training images captured at one or more focal positions. The one or more processors may further receive one or more target feature selections for one or more target overlay measurements corresponding to one or more target features of a target specimen. The one or more processors may further determine one or more target focal positions based on the one or more target feature selections using the machine learning classifier. The one or more processors may further receive one or more target images captured at the one or more target focal positions, the one or more target images including the ...

THREAT ANALYSIS AND RISK ASSESSMENT FOR CYBER-PHYSICAL SYSTEMS BASED ON PHYSICAL ARCHITECTURE AND ASSET-CENTRIC THREAT MODELING

СИСТЕМА АНАЛИЗА ИНТЕРНЕТ-ТРАФИКА ПОЛЬЗОВАТЕЛЯ ДЛЯ ПРИСВОЕНИЯ ПОЛЬЗОВАТЕЛЬСКОМУ УСТРОЙСТВУ КОЭФФИЦИЕНТА КЛАССИФИКАЦИИ ПОЛЬЗОВАТЕЛЯ

Изобретение относится к вычислительной технике. Технический результат заключается в повышении точности классификации пользователя. Система анализа Интернет-трафика пользователя для присвоения пользовательскому устройству коэффициента классификации пользователя содержит сервер, выполненный с возможностью: установки интернет-соединения с вычислительным устройством пользователя для анализа аппаратных средств вычислительного устройства пользователя с присвоением вычислительному устройству пользователя коэффициента аппаратных средств вычислительного устройства пользователя; установки интернет-соединения с вычислительным устройством пользователя для анализа параметров интернет-соединения с присвоением вычислительному устройству пользователя коэффициента интернет-соединения вычислительного устройства пользователя; установки интернет-соединения с вычислительным устройством пользователя для анализа поведения пользователя в Интернет-сессии с присвоением вычислительному устройству пользователя поведенческого ...

Способ классификации текстовой информации электронного вида на предмет наличия конфиденциальных данных

Заявленное изобретение относится к области вычислительной техники для выявления текстовой информации электронного вида, содержащей конфиденциальные данные. Заявленный способ заключается в том, что после этапа предварительной обработки и векторизации корпуса текстовой информации осуществляется процедура разделения выборки на тестовую и обучающую. Затем происходит последовательное обучение моделей классификации. Обученные модели тестируются, производится расчет значений их точности, а результаты тестирования комбинируются с целью исправления ошибок первого и второго рода при осуществлении классификации отдельно взятыми моделями классификации текстовой информации электронного вида на предмет наличия в них конфиденциальных данных. Далее рассчитывается значение показателей точности для каждой сформированной комбинации моделей, среди которых выбирается наибольшее значение точности. Техническим результатом является повышение точности классификации текстовой информации электронного вида по степени ...

Method and apparatus for determining encryption mask, device and storage medium

Encryption mask determination comprises acquiring (201) test image and encryption mask sets and superimposing (202) an image in the image set with a mask in the mask set to obtain an encrypted image set. An image in the encrypted image set is recognised (203) using pre-trained encrypted and original image recognition models, both providing a recognition result. A target encryption mask is determined (204) from the mask set based on the result. As the target mask is determined from the set according to both pre-trained models, image security is improved due to recognition precision. Also claimed is training a model (Fig. 9), comprising acquiring 1st image (1st training sample) and encryption mask sets and superimposing a 1st set image with a randomly sampled mask to obtain a 2nd training sample. A 2nd image set is acquired (3rd training sample). Training a 1st initial model based on the 1st sample provides an original image recognition model. Training a 2nd initial model based on the 2nd ...

Model training method and apparatus, and prediction method and apparatus

Exposure defects classification of images using a neural network

A method 300 of training a neural network to classify image exposure defects comprises: receiving a first set 302 of labelled images; training a neural network 304 using the labelled images; receiving a second set 306 of unlabelled images; applying the second set of images to the neural network to produce inference scores 308 based on the exposure quality; classifying each image of the second set as well-exposed or poorly exposed based on the inference scores; and retraining the neural network model based on the classifying and the second set of images. The first set of images are each labelled as either well-exposed or defective. The classifying is carried out based on a distribution ratio of well-exposed and defective images in the first set in addition to the inference scores. The defects may be underexposure or overexposure. The retraining may comprise determination, based on the distribution ratio and inference scores, that an image of the second set is misclassified, and retraining ...

System and method for performing object analysis

There is provided a system that, when in operation, classifies one or more elements within an input signal, wherein the system: receives a compressed version of the input signal, wherein the compressed version comprises at least two sets of compressed data in a hierarchy, wherein each set of compressed data, when decoded, enables the signal to be reconstructed up to a respective level of quality; decodes the compressed version of the signal up to a first level of quality by decoding a first set of compressed data to generate a first reconstructed signal; performs a first classification operation on the first reconstructed signal; decodes the compressed version of the signal up to a second level of quality by decoding a second set of compressed data to generate a second reconstructed signal; and performs one or more second classification operations on the second reconstructed signal.

Video data processing

A method comprising receiving raw video data, detecting regions of interest in a frame of the video data, performing a cropping process on the video data, performing demosaicing on the cropped video data to generate processed image data, and, analysing the processed image data to determine information related to the regions of interest. The ROIs may be detected using a first neural network and the image data may be analysed using a second neural network. The ROIs may be determined based on an object detection process using a region proposal network. Processing resources applied to the method may be scaled depending on the size of the ROIs.

Systems and methods for detecting symptoms of occupant illness

Method of detecting symptoms of illness, comprising: receiving an image 602, audio data 402 and/or radar data 1102; classifying portions of the image, audio data and/or radar data as indicating an illness symptom, such as a cough or sneeze 610, 408, 1110; determining a location of the recognized symptom utilizing the image data from the image source 612, 412, 1112; and displaying an indicator that a recognized symptom occurred at a location 1508. The audio data may come from a plurality of audio sources and the symptoms’ location 412 determined through triangulation. The RADAR data may be used to detect a person’s vital signs. Where more than one data source is used, fusion of the data may occur prior (1502, Fig. 15) or subsequent to 1602 classifying an activity from the data (1110, Fig. 15; 410, 610, 1110, Fig. 16). Recognised symptoms may be aggregated (summed) over time 1506 to provide a color-coded frequency heat map (500, Fig. 5) indicating the location of the recognised symptoms ( ...

Acoustic source classification using hyperset of fused voice biometric and spatial features

Systems and methods for automated detection of changes in extent of structures using imagery

Retail checkout terminal fresh produce identification system

Disclosed are systems and methods including starting with a first number of images, generating a second number of images by digital operations on the first number of images, extracting features from the second number of images, and generating a classification model by training a neural network on the second number of images wherein the classification model provides a percentage likelihood of an image's categorisation, embedding the classification model in a processor and receiving an image for categorisation, wherein the processor is in communication with a POS system, the processor running the classification model to provide output to the POS system of a percentage likelihood of the image's categorisation.

AVIAN DETECTION SYSTEMS AND METHODS

Provided herein are detection systems and related methods for detecting moving objects in an airspace surrounding the detection system. In an aspect, the moving object is a flying animal and the detection system comprises a first imager and a second imager that determines position of the moving object and for moving objects within a user selected distance from the system the system determines whether the moving object is a flying animal, such as a bird or bat. The systems and methods are compatible with wind turbines to identify avian(s) of interest in airspace around wind turbines and, if necessary, take action to minimize avian strike by a wind turbine blade.

Single sign-on through customer authentication systems

Described herein is a system, method, and non-transitory computer readable medium related to a service provider using a third party identity provider to authenticate a user with improved security. An authentication token is received from the identity provider, and can be verified against internal configuration information. The internal configuration information includes data that is not included in the authentication token, and therefore, is not vulnerable to some security attacks, such as a man-in-the-middle attack. After the authentication token is verified, the internal configuration information and authentication token may be used to create a custom identifier, referred to as an identity ID. The identity ID may be used by the service provider to verify user access to resources.

Wafer map analyzer, method for analyzing wafer map using the same and method for manufacturing semiconductor device

A method for analyzing a wafer map using a wafer map analyzer includes generating first wafer maps each displaying characteristics of a first wafer for a corresponding channel of a plurality of channels. The first wafer maps are auto-encoded together to extract a first feature. The method also includes determining whether the first feature is a valid pattern, classifying the type of the first feature based on unsupervised learning when the first feature is a valid pattern and extracting a representative image of features classified into the same type as the first feature.

Method and system for determining response for digital task executed in computer-implemented crowd-sourced environment

Disclosed are a method and a system for determining a response to a digital task in a computer-implemented crowd-sourced environment. The method comprises determining if a number of the plurality of responses to the digital task received meets a pre-determined minimum answer threshold; in response to the number of the plurality of responses to the digital task meeting the pre-determined minimum answer threshold, executing: for each of the plurality of responses generating, by the server, a confidence parameter representing a probability of an associated one of the plurality of responses being correct; ranking the plurality of responses based on the confidence parameter to determine a top response being associated with a highest confidence parameter; and in response to the highest confidence parameter being above a pre-determined minimum confidence threshold, assigning a value of the top response as a label for the digital task and terminating the digital task execution.

Recognizing the movement intention of a pedestrian from camera images

A method and a driver assistance system recognize the intention of a pedestrian to move on the basis of a sequence of images of a camera. The method includes detecting a pedestrian in at least one camera image. The method also includes selecting a camera image that is current at the time t and selecting a predefined selection pattern of previous camera images of the image sequence. The method further includes extracting the image region in which the pedestrian was detected in the selected camera images of the image sequence. The method also includes classifying the movement profile of the detected pedestrian on the basis of the plurality of extracted image regions. The method outputs the class that describes the movement intention determined from the camera images of the image sequence.

Logo recognition in images and videos

Accurately detection of logos in media content on media presentation devices is addressed. Logos and products are detected in media content produced in retail deployments using a camera. Logo recognition uses saliency analysis, segmentation techniques, and stroke analysis to segment likely logo regions. Logo recognition may suitably employ feature extraction, signature representation, and logo matching. These three approaches make use of neural network based classification and optical character recognition (OCR). One method for OCR recognizes individual characters then performs string matching. Another OCR method uses segment level character recognition with N-gram matching. Synthetic image generation for training of a neural net classifier and utilizing transfer learning features of neural networks are employed to support fast addition of new logos for recognition.

METHOD, APPARATUS, AND SYSTEM FOR PROVIDING LINEAR FEATURE DETECTION OF CARTOGRAPHIC FEATURES

An approach is provided for linear feature detection of cartographic features. The approach, for example, involves receiving a plurality of linear feature detections that represent one or more linear features of a geographic environment and are detected using at least one sensor. The approach also involves clustering the plurality of linear feature detections into at least one cluster and determining that the at least one cluster forms a polygon. The approach further involves cutting the polygon at one or more cut locations to form a plurality of sub-clusters of the plurality of linear feature detections. The approach further involves extracting respective centerlines for the plurality of sub-clusters and connecting the respective centerlines at the one or more cut locations to generate a network graph of the one or more linear features.

METHODS AND APPARATUS FOR ACTIONS, ACTIVITIES AND TASKS CLASSIFICATIONS BASED ON MACHINE LEARNING TECHNIQUES

CONNECTED DEVICE REGION IDENTIFICATION

Various aspects of methods, systems, and use cases include region identification of a malicious device based on crowdsourced locations. A method may include generating a grid of bins of a local radio coverage region, receiving location data from a plurality of devices in the local radio coverage region, and classifying locations of the plurality of devices with respect to the bins. The method may include associating the classified locations of the plurality of devices to the received location data for corresponding devices of the plurality of devices, and generating a model, from the associated classified locations and the received location data.

СПОСОБ ГРАФОВОЙ НЕЙРОСЕТЕВОЙ КЛАССИФИКАЦИИ НА ОТСУТСТВИЕ ИЛИ НАЛИЧИЕ БОЛЬШОГО ДЕПРЕССИВНОГО РАССТРОЙСТВА ПО ДАННЫМ ФМРТ

Изобретение относится к способу графовой нейросетевой классификации на отсутствие или наличие большого депрессивного расстройства по данным фМРТ. Технический результат заключается в повышении точности определения наличия большого депрессивного расстройства по данным фМРТ. В способе регистрируют сигнал функциональной МРТ головного мозга в состоянии покоя у испытуемых контрольной группы и испытуемых с диагностированным большим депрессивным расстройством, в результате чего строят, в соответствии с автоматическим атласом анатомической разметки AAL3, корреляционную матрицу размером 166*166, каждый элемент корреляционной матрицы хранит в себе коэффициент корреляции Пирсона, к элементам матрицы применяют 20 пороговых значений в диапазоне от 0,0 до 0,95 с шагом 0,05, при этом значения коэффициента корреляции ниже порогового значения зануляют и из исходной матрицы получают 20 наборов матриц разной степени разреженности, бинаризуют матрицы путем присвоения значения «1» для коэффициентов корреляции ...

Knowledge distillation-based compression method for pre-trained language model, and platform

A knowledge distillation-based compression method for a pre-trained language model, and a platform. In the method, a universal feature transfer knowledge distillation strategy is first designed, and in a process of distilling knowledge from a teacher model to a student model, feature maps of each layer of the student model are approximated to features of the teacher model, with emphasis on the feature expression capacity in intermediate layers of the teacher model for small samples, and these features are used to guide the student model; then, the ability of the self-attention distribution of the teacher model to detect semantics and syntax between words is used to construct a knowledge distillation method based on self-attention crossover; and finally, in order to improve the learning quality of early-period training and the generalization ability of late-period training in the learning model, a Bernouli probability distribution-based linear transfer strategy is designed to gradually complete ...

Force sensor sample classification

A classification system for classifying sensor samples in a sensor system, the sensor system comprising N force sensors each configured to output a sensor signal, where N≥1, each sensor sample comprising N sample values from the N sensor signals, respectively, the classification system comprising: a classifier configured, for each of a series of candidate sensor samples, to perform a classification operation based on the N sample values concerned and generate a classification result which labels the candidate sensor sample as indicative of a defined target event, thereby generating a series of classification results corresponding to the series of candidate sensor samples, respectively; a determiner configured to output at least one event determination based on the series of classification results; and a controller configured to control the classifier and/or the determiner based on one or more controller input signals.

Neural network training using robust temporal ensembling

패션 스타일 분류 시스템 및 그 분류 방법

... 패션 스타일 분류 시스템은, 패션 객체의 이미지가 포함된 상품 사진으로부터 상기 패션 객체의 이미지를 추출하는 패션 객체 이미지 추출기; 상기 패션 객체 이미지 추출기로부터 추출된 상기 패션 객체의 이미지로부터, 상기 패션 객체에 속하는 다수의 구성 부분 이미지를 추출하는 구성 부분 추출기; 및 상기 구성 부분 추출기로부터 추출된 다수의 구성 부분 이미지 각각에 대한 표준 이미지를 생성하는 구성 부분 표준 이미지 생성기;를 포함한다.

METHOD AND SYSTEM TO REDUCE LATENCY OF TOUCH EVENTS

... 시스템은: 복수의 터치 이벤트들을 수신하고 터치 이벤트들에 기초하여 출력을 생성하기 위한 터치 경로 모듈; 및 비디오 이미지를 수신하고; 터치 경로 모듈의 출력을 수신하고; 형상의 경계를 정의하는 복수의 포인트들을 결정하고 ―형상은 터치 경로 모듈의 출력에 기초하여 결정됨―; 결합된 디스플레이 이미지를 생성하기 위해 복수의 포인트들에 따라 비디오 이미지를 오버레이 데이터와 결합하고; 결합된 디스플레이 이미지를 출력하기 위한 렌더링 모듈을 포함한다.

INFORMATION PROCESSING METHOD, STORAGE MEDIUM, AND INFORMATION PROCESSING APPARATUS

It is intended to provide a significant data expansion algorithm for predetermined data. An information processing method performed by a processor included in an information processing device, the method includes: acquiring expanded data resulting from expansion of target data using an optional data expansion algorithm including a coupled function obtained by coupling together a plurality of data expandable functions by using weights; implementing learning, the learning including implementing the learning by inputting the expanded data to a learning model that performs predetermined learning and implementing the learning by using each item of the expanded data generated by stepwise changing a weight of the coupled function; specifying a boundary weight with which a learning result of the learning indicates an intended result and associating the boundary weight with information related to the target data.

Methods, systems, articles of manufacture and apparatus to manage process excursions

Methods, apparatus, systems and articles of manufacture are disclosed to manage process excursions. An example apparatus includes a digital twin comparer to determine when a product fails to satisfy a tolerance metric of a digital twin, and a fingerprint manager to generate a fingerprint corresponding to a sensor pattern. The example apparatus also includes a node interfacer to determine a number of workstations of a process control system that exhibit the fingerprint, and an excursion statistics calculator to invoke a corrective action for respective ones of the number of workstations, the corrective action based on a threshold count of the number of workstations that exhibit the fingerprint.

Systems and methods for time series analysis using attention models

A system for time series analysis using attention models is disclosed. The system may capture dependencies across different variables through input embedding and may map the order of a sample appearance to a randomized lookup table via positional encoding. The system may capture capturing dependencies within a single sequence through a self-attention mechanism and determine a range of dependency to consider for each position being analyzed. The system may obtain an attention weighting to other positions in the sequence through computation of an inner product and utilize the attention weighting to acquire a vector representation for a position and mask the sequence to enable causality. The system may employ a dense interpolation technique for encoding partial temporal ordering to obtain a single vector representation and a linear layer to obtain logits from the single vector representation. The system may use a type dependent final prediction layer.

SYSTEMS AND METHODS FOR ARTIFICIAL INTELLIGENCE (AI) ERGONOMIC POSITIONING

An Artificial Intelligence (AI) ergonomic assessment and positioning system that analyzes remote workspace data, identifies objects that are improperly positioned, oriented, and/or have undesirable settings, and automatically adjusts, moves, sets, and/or provides automatic guidance for the adjustment, movement, and/or setting of target objects in the remote workspace.

Method and system for face detection

A method for identifying one or more faces in an image may include determining a detection region in an image based on at least one of a plurality of detection scales, and identifying one or more faces in the detection region based on the at least one of the plurality of detection scales. The method may further include calibrating the detection region based on one or more identified faces.

Defect detection of a component in an assembly

A system for validating installation correctness of a component in a test assembly includes a housing having a platform including a tiered surface. The tiered surface forms an abutment surface configured as a stop against which a test assembly is abutted. A plurality of cameras is positioned to capture different views of the test assembly. A processing device is configured to execute instructions to capture an image from each of the plurality of cameras of the test assembly that includes a plurality of components. Each of the plurality of components is analyzed within each image of the plurality of images. A matching score is determined and an indication of whether the plurality of components was correctly installed in the test assembly is generated.

Image based counterfeit detection

Systems and methods for authenticating material samples are provided. Digital images of the samples are processed to extract computer-vision features, which are used to train a classification algorithm along with location and optional time information. The extracted features/information of a test sample are evaluated by the trained classification algorithm to identify the test sample. The results of the evaluation are used to track and locate counterfeits or authentic products.

Digital watermarking of machine-learning models

In one embodiment, a method includes by a computing device, detecting a sensory input, identifying, using a machine-learning model, one or more attributes associated with the machine-learning model, wherein the attributes are identified based on the sensory input in accordance with the model's training, and presenting the attributes as output. The identifying may be performed at least in part by an inference engine that interacts with the model. The sensory input may include an input image received from a camera, and the model may identify the attributes based on an input object in the input image in accordance with the model's training. The model may include a convolutional neural network trained using training data that associates training sensory input with the attributes. The training sensory input may include a training image of a training object, and the input object may be classified in the same class as the training object.

METHODS AND SYSTEMS FOR DETECTING CRACKS IN ILLUMINATED ELECTRONIC DEVICE SCREENS

Systems and methods for detecting the cracks in illuminated electronic device screens are disclosed. In one embodiment, the method includes receiving an image of an electronic device screen and retrieving a plurality of kernels, each having values corresponding to a line region and a non-line region, with the orientation of the line region and the non-line region differing for each kernel. At least some of the kernels are applied to the image to obtain, at various locations of the image, values corresponding to the line regions and the non-line regions. Based on the values corresponding to the line regions and the non-line regions, cracks are automatically identified in the electronic device screen.

Labeling images using a neural network

Apparatuses, systems, and techniques to generate labels for images using generative adversarial networks. In at least one embodiment, one or more objects in an input image are identified using one or more generative adversarial networks (GANs) and a synthetic version of the input image and one or more labels corresponding to the one or more objects within the synthetic version of the input image are generated using the GANs.

Augmented reality method and system for design

A system for generating formal premises designs is discussed. The system begins by inputting an informal two-dimensional rendering of the premises. The system continues by analyzing the informal two-dimensional rendering and determining features found on the premises depicted by the two-dimensional rendering. The system then generates a three-dimensional model of the premises and populates it with interactive objects. Finally, the system provides multiple views of the three-dimensional rendering containing interactive objects.

결함 분류를 위한 방법 및 시스템

... 결함 분류는, 시료의 하나 이상의 이미지를 획득하는 것, 하나 이상의 트레이닝 결함의 하나 이상의 속성에 기초하여 하나 이상의 트레이닝 결함의 수동 분류를 수신하는 것, 수신된 수동 분류 및 하나 이상의 트레이닝 결함의 속성에 기초하여 앙상블 학습 분류자를 생성하는 것, 수신된 분류 순도 요건에 기초하여 하나 이상의 트레이닝 결함의 각각의 결함 타입에 대한 신뢰도 임계치를 생성하는 것, 하나 이상의 테스트 결함을 포함하는 하나 이상의 이미지를 획득하는 것, 하나 이상의 테스트 결함을 생성된 앙상블 학습 분류자로 분류하는 것, 하나 이상의 테스트 결함의 각각에 대한 신뢰도 레벨을 생성된 앙상블 학습 분류자로 계산하는 것 및 생성된 신뢰도 임계치 아래의 신뢰도 레벨을 갖는 하나 이상의 테스트 결함을 수동 분류를 위해 유저 인터페이스 디바이스를 통해 보고하는 것을 포함한다.

로봇 행동 보정에 기초한 로컬 피쳐 모델의 업데이트

... 로봇 행동에 대한 보정을 결정하고 이용하기 위한 방법, 장치, 및 컴퓨터 판독가능한 매체가 개시된다. 일부 구현들은, 로봇에 의해 수행되는 행동의 인간 보정을 결정하는 것에 응답하여, 로봇의 로컬 피쳐 모델(feature model)을 업데이트하는 것에 관한 것이다. 로컬 피쳐 모델은, 대응하는 신경망 모델을 통해 생성된 임베딩(embedding)에 기초하여, 생성된 임베딩과 가장 유사한 하나 이상의 피쳐를 결정하는데 이용된다. 인간 보정에 응답하여 로컬 피쳐 모델을 업데이트하는 것은, 인간 보정에 대응하는, 로컬 피쳐 모델의 피쳐 임베딩을 업데이트하는 것을 포함할 수 있다. 피쳐 모델에 대한 조정(들)은 대응하는 신경망 모델의 재훈련을 필요로하지 않고도 로봇 수행을 즉시 향상시킬 수 있다.

전이 학습을 이용한 이미지 분류 모델 생성 장치 및 방법

... 본 발명의 일 실시예는, 이미지 분류 모델 생성 장치를 이용하여 이미지 분류 모델을 생성하는 방법을 제공한다. 본 방법은, 학습에 사용되는 이미지 데이터를 대상으로 전처리를 수행하여 전처리 데이터를 생성하는 단계, 상기 전처리 데이터를 대상으로 데이터 확대를 수행하여 학습 데이터 세트를 생성하는 단계, 그리고, 상기 이미지 데이터와 상이한 이미지 데이터를 대상으로 분류 학습을 수행하여 생성된 분류 학습 모델들 중에서 하나의 분류 학습 모델을 선택하고, 선택된 분류 학습 모델을 이용하여 상기 학습 데이터 세트를 대상으로 하는 전이 학습을 수행하는 단계를 포함한다. 상기 전이 학습은, 상기 선택된 분류 학습 모델에 포함된 분류기와 복수개의 학습 레이어들을 분리시켜 상기 학습 데이터 세트를 대상으로 분류 학습을 수행하도록 하는 것을 의미한다.

딥러닝 기반 의상 속성 분류 장치 및 방법

... 딥러닝 기반 의상 속성 분류 장치 및 방법이 개시된다. 본 발명의 실시예에 따른 딥러닝 기반 의상 속성 분류 장치는, 적어도 하나의 프로그램이 기록된 메모리 및 프로그램을 실행하는 프로세서를 포함하며, 프로그램은, 입력 이미지에 포함된 사람이 착용한 의상에 대한 적어도 하나 이상의 속성에 대한 제1 분류 결과를 출력하는 제1 분류부, 입력 이미지에 포함된 사람의 신체를 분해한 주요 부위 영역들 각각에 상응하는 복수의 마스크 레이어들이 적층된 마스크 텐서를 출력하는 마스크 생성부, 마스크 생성부에 의해 생성된 마스크 텐서를 적용하여 입력 이미지에 포함된 사람이 착용한 의상에 대한 적어도 하나 이상의 속성에 대한 제2 분류 결과를 출력하는 제2 분류부 및 제1 분류 결과 및 제2 분류 결과를 기반으로 입력 이미지에 대한 최종 분류 결과를 결정하여 출력하는 최종 분류부를 포함할 수 있다.

클라이언트 애플리케이션 콘텐츠 분류 및 발견

... 하나 이상의 구현에서, 클라이언트 애플리케이션을 사용하여 생성된 콘텐츠는 분류와 연관될 수 있다. 콘텐츠는 콘텐츠와 연관된 하나 이상의 분류에 기초하여 액세스될 수 있다. 콘텐츠는 콘텐츠 생성자의 클라이언트 디바이스에 의해 실행되는 클라이언트 애플리케이션에 의해 생성되어 콘텐츠의 수신자들의 하나 이상의 추가적인 클라이언트 디바이스에 콘텐츠를 배포할 수 있는 서버 시스템에 전송될 수 있다. 서버 시스템은 또한 동일한 분류를 갖는 콘텐츠를 모을 수 있다.

변화 검출 기반의 이미지 획득 시스템

... 이미지 획득 시스템을 태스킹하는 시스템 및 방법이 제공된다.일 실시예에서, 방법은 지리적 영역과 관련된 복수의 이미지를 기술하는 데이터를 획득하는 단계를 포함한다. 상기 방법은 상기 지리적 영역과 관련된 하나 이상의 변화의 발생을 결정하기 위해 상기 복수의 이미지의 적어도 하나의 서브 세트를 분석하는 단계를 포함한다. 상기 방법은 지리적 영역과 관련된 하나 이상의 변화의 발생에 적어도 부분적으로 기초하여 지리적 영역과 관련된 변화 레벨을 결정하는 단계를 포함한다. 이 방법은, 변화 레벨에 적어도 부분적으로 기초하여 지리적 영역과 관련된 영상 데이터의 획득을 조정하도록 영상 획득 시스템에 제어 명령을 제공하는 단계를 포함한다.

ELECTRONIC APPARATUS AND OPERATION METHOD THEREOF

Provided are a method for recognizing a disposition type of an article based on attribute information of an article, and updating information on inventory based on the disposition information of an article accompanying when the article is processed according to the disposition type, and an electronic device therefor.

APPARATUS AND METHOD FOR DETECTING ANIMAL SPECIES

According to one embodiment of the present invention, provided is an animal species recognition device, which comprises: a group generation unit which generates a group of input images in accordance with color information; a data set generation unit which generates a data set including the input image of the group and a database image and a label corresponding to the group; a training unit which trains the Siamese Network in accordance with the data set; and a recognition unit which generates species recognition information corresponding to a target image by inputting the target image to the Siamese network.

Multi-model medical scan analysis system using fine-tuned models

A multi-model medical scan analysis system is operable to generate a generic model by performing a training step on image data of a plurality of medical scans and corresponding labeling data. A plurality of fine-tuned models are generated by performing a fine-tuning step on the generic model. Abnormality detection data is generated for a new medical scan by utilizing the generic model. A first one of the plurality of abnormality types that is detected in the new medical scan is determined based on a corresponding one of the plurality of probability values. Additional abnormality data is generated by performing a fine-tuned inference function on the image data of the new medical scan that utilizes one of the plurality of fine-tuned models that corresponds to the first one of the plurality of abnormality types. The additional abnormality data is transmitted for display.

EDGE METADATA

In one embodiment, a method includes receiving, by a storage device located at an edge of a network, user device data from a first user device and receiving, by the storage device, metadata associated with an environment of the first user device. The method also includes classifying, by the storage device, the metadata into a category to generate classified metadata and applying, by the storage device, a policy to the classified metadata to generate policy-based classified metadata. The method further includes assigning, by the storage device, one or more tokens to the policy-based classified metadata and assigning, by the storage device, one or more tags to the policy-based classified metadata. The method further includes transmitting, by the storage device, the one or more tags to a second user device in accordance with the one or more tokens.

COMPUTER-BASED SYSTEMS HAVING DATA STRUCTURES CONFIGURED FOR MACHINE LEARNING CLASSIFICATION OF ENTITIES AND METHODS OF USE THEREOF

At least some embodiments are directed to an entity classification system receives informational data associated with an entity. The informational data includes sentences associated with the entity. The entity classification system utilizes a first machine learning model to determine a first contextual meaning among words of a sentence associated with the entity based on a first word embedding technique, and determines at least one category associated with the entity based at least in part on the first contextual meaning. The entity classification system utilizes a second machine learning model to determine a second contextual meaning shared by a set of sentences based on a second embedding technique, and determines a subcategory of the category associated with the entity based at least in part on the second contextual meaning. The entity classification system generates an output including the category and subcategory associated with the entity.

Multi-modal data explainer pipeline

Embodiments of the present invention provide a computer system, a computer program product, and a method that comprises ingesting tabular data from at least one modality of a plurality of modalities; simultaneously extracting data and generating a prediction model for a task of a computing device from the extracted data from at least two modalities in the plurality of modalities; generating a data signature based on the generated prediction model from the at least two modalities by leveraging the generated prediction model for ingested tabular data and extracted data; comparing the generated data signature to identified data signatures stored in at least one modality in the plurality of modalities; and performing a task based on the generated data signature and a validation of the comparison of identified data signatures.

SYSTEMS AND METHODS TO IDENTIFY NEURAL NETWORK BRITTLENESS BASED ON SAMPLE DATA AND SEED GENERATION

Systems and methods for determining neural network brittleness are disclosed. For example, the system may include one or more memory units storing instructions and one or more processors configured to execute the instructions to perform operations. The operations may include receiving a modeling request comprising a preliminary model and a dataset. The operations may include determining a preliminary brittleness score of the preliminary model. The operations may include identifying a reference model and determining a reference brittleness score of the reference model. The operations may include comparing the preliminary brittleness score to the reference brittleness score and generating a preferred model based on the comparison. The operations may include providing the preferred model.

Systems and techniques for securely processing disparate data sets in spreadsheets

An aspect of the present teachings generally includes a data processing platform (e.g., a platform hosted by a remote computing resource) that analyzes and compiles information contained in a source spreadsheet, e.g., to ascertain how the source spreadsheet manipulates input data. The data processing platform may create an encrypted file related to the source spreadsheet that includes transformed input data and/or directives that define functionality of the source spreadsheet, e.g., where the encrypted file is specifically tailored for use on the data processing platform for a particular user or set of users. Also, or instead, the data processing platform may create or otherwise utilize one or more runtimes that, when executed in conjunction with the encrypted file, can mimic functionality of the source spreadsheet and/or other useful functionality applied to the input data or other disparate data.

Systems and methods for managing meat cut quality

In some embodiments, apparatuses and methods are provided herein useful to ensuring quality of meat cuts. In some embodiments, a system for ensuring quality of meat cuts comprises a capture device comprising an image capture device configured to capture an image of a cut of meat, a depth sensor configured to capture depth data, a transceiver configured to transmit the image and the depth data, a microcontroller configured to control the image capture device, the depth sensor, and the transceiver, a database configured to store meat cut specifications, and the control circuit configured to receive, from the capture device, the image and the depth data, retrieve, from the database, a meat cut specification, evaluate the image of the cut of meat and the depth data associated with the cut of meat, and classify the cut of meat.

Camera parameter estimation using semantic labels

In one implementation, a method of determining an estimate of a position of a camera is performed at a device including one or more processors and non-transitory memory. The method includes obtaining a point cloud of a physical environment including a plurality of points, wherein each of the plurality of points is associated with set of three-dimensional coordinates in a three-dimensional coordinate system of the physical environment, wherein the plurality of points includes a first cluster of points associated with a first semantic label. The method includes obtaining a two-dimensional image of the physical environment associated with a point-of-view set of three-dimensional coordinates in the three-dimensional coordinate system of the physical environment. The method includes detecting, in the two-dimensional image of the physical environment, a representation of a first object corresponding to the first semantic label. The method includes determining a first plurality of sets of two-dimensional ...

MONITORING DEVICE

LEARNING AND PREDICTING TEMPORAL PROFILES OF PHYSIOLOGICAL STATES ASSOCIATED WITH THE ADMINISTRATION OF COMMONLY USED CRITICAL CARE DRUGS

АВТОМАТИЗИРОВАННАЯ СИСТЕМА ПРОСМОТРА КОНТЕНТА С ВЫСОКИМ ДИНАМИЧЕСКИМ ДИАПАЗОНОМ В РЕАЛЬНОМ ВРЕМЕНИ

Изобретение относится к области обработки изображений. Техническим результатом является повышение качества отображения изображений. Способ регулировки яркости изображения и выполнения рендеринга изображения начинается с того, что в изображении идентифицируется светлая область. Светлая область классифицируется в присвоенную классификацию. Значение яркости светлой области определяется и сравнивается с заранее заданными значениями яркости, соответствующими классификации. Значение яркости светлой области регулируется для согласования с заранее заданными значениями яркости при наличии рассогласования. Выполняется рендеринг изображения с идентифицированной светлой областью, имеющей отрегулированное значение яркости. 3 н. и 12 з.п. ф-лы, 10 ил.

СПОСОБ И СИСТЕМА ОБЕЗЛИЧИВАНИЯ КОНФИДЕНЦИАЛЬНЫХ ДАННЫХ

Изобретение относится к области защиты данных, а более конкретно к обратимому обезличиванию конфиденциальных данных с сохранением структуры данных в текстовых документах. Техническим результатом является обеспечение возможности сохранения стилистической, семантической, лексической и морфологической структуры данных в текстовых документах при их обезличивании, повышение точности обезличивания данных в текстовых документах за счет определения конфиденциальных данных в текстовых документах с помощью модели машинного обучения. Способ обратимого обезличивания конфиденциальных данных в текстовых документах с сохранением структуры данных содержит этапы, на которых: получают документ с текстовыми данными; сегментируют и токенизируют текстовые данные, выполняют векторизацию токенов; определяют принадлежности каждого токена к категории конфиденциальных данных с помощью модели машинного обучения; обезличивают данные, относящиеся к токенам с конфиденциальными данными с сохранением структуры данных; ...

System and method for generating user inputs for a video game

Video data processing

소셜 미디어 데이터에 대한 준 지도 멀티 모달 딥 임베디드 클러스터링을 이용한 인간 활동 인식 방법

... 본 발명의 일 실시예에 따른 소셜 미디어 데이터에 대한 준 지도 멀티 모달 딥 임베디드 클러스터링을 이용한 인간 활동 인식 방법은, 훈련 데이터로 제공된 이미지 및 텍스트를 포함하는 멀티 모달 소셜 미디어 데이터를 전처리하여 이미지 특징 및 텍스트 특징을 추출하는 멀티 모달 데이터 전처리 단계; 상기 이미지 특징에 대한 이미지 임베딩 네트워크 및 상기 텍스트 특징에 대한 텍스트 임베딩 네트워크의 초기화를 수행하는 임베딩 네트워크 초기화 단계; 상기 훈련 데이터 중 레이블이 있는 데이터를 사용하여 상기 이미지 임베딩 네트워크 및 텍스트 임베딩 네트워크를 학습하는 지도 학습 단계; 및 상기 훈련 데이터에 포함된 레이블이 없는 데이터 및 레이블이 있는 데이터를 모두 사용하여 상기 이미지 임베딩 네트워크 및 텍스트 임베딩 네트워크를 학습하는 비지도 학습 단계를 포함할 수 있다.

복수의 이상 탐지 기법을 통합하여 공정 설비의 이상을 탐지하는 방법, 장치 및 프로그램

... 본 발명의 다양한 실시예에 따른 복수의 이상 탐지 기법을 통합하여 공정 설비의 이상을 탐지하는 방법이 개시된다. 상기 방법은: 공정 설비에서 수집되는 Bin 데이터, PCM(Process Control Monitoring) 데이터, 복수의 웨이퍼맵 및 FDC(Fault Detection and Classification) 데이터를 수집하는 단계; 상기 Bin 데이터를 기초로 불량률이 높은 순서대로 기 설정된 개수의 특정 Bin 데이터를 추출하는 단계; 상기 PCM 데이터를 기초로 스펙 불량 또는 컨트롤 불량이 발생된 특정 소자를 추출하는 단계; 상기 복수의 웨이퍼맵을 기초로 상기 복수의 웨이퍼맵 각각의 패턴을 분류하고, 분류된 패턴에 대응하는 적어도 하나의 특정 불량 장비를 추출하는 단계; 상기 FDC 데이터를 기초로 상기 공정 설비의 비정상 동작에 대한 불량 여부를 추출하는 단계; 및 상기 Bin 데이터, 상기 PCM 데이터, 상기 복수의 웨이퍼맵 및 상기 FDC 데이터 각각을 기초로 추출된 결과를 통합하여 제공하는 단계;를 포함할 수 있다.

데이터 시각화 기반 인터렉티브 미디어 아트 제공 플랫폼 운영 서버 및 방법

... 본 발명의 일 측면은, 데이터 시각화 기반 인터렉티브 미디어 아트 제공 플랫폼 운영 서버를 제공한다. 데이터 시각화 기반 인터렉티브 미디어 아트 제공 플랫폼 운영 서버는, 적어도 하나의 프로세서(processor) 및 적어도 하나의 프로세서가 적어도 하나의 단계를 수행하도록 지시하는 명령어들(instructions)을 저장하는 메모리(memory)를 포함할 수 있다. 적어도 하나의 단계는, 데이터를 수신하여 데이터 정보를 추출하고, 데이터 정보를 속성에 따라 적어도 하나의 카테고리로 분류하는 단계, 작품 정보를 수신하여, 작품 정보를 속성에 따라 적어도 하나의 카테고리로 분류하는 단계, 분류된 각 카테고리의 데이터 정보를 상호 결합하여, 데이터 정보 점수를 산출하는 단계, 분류된 각 카테고리의 작품 정보를 상호 결합하여, 작품 정보 점수를 산출하는 단계 및 데이터 정보 점수 및 작품 정보 점수를 기초로 수신한 작품을 변경하여 데이터 시각화를 수행하는 단계를 포함할 수 있다.

인공지능 기반 안전 영상 판단 시스템

... 본 발명은 인공지능 기반 안전 영상 판단 시스템에 관한 것이다. 구체적으로, 미리 설정된 레퍼런스 이미지 정보에 기반하여, 특정 촬영 구간의 객체 정보 및 상황 정보를 포함하는 제 1 이미지 획득 모듈; 상기 획득된 이미지의 품질 향상을 위하여, 머신 러닝 기반의 전처리(pre-processing)를 수행하여상기 제 1 이미지로에 비하여 이미지 품질이 향상된 제 2 이미지를 획득하는 전처리(pre-processing) 모듈; 상기 제 2 이미지 내부에 포함된 객체 정보 및 상황 정보를 상기 선택적 분석 처리를 수행하여 상기 상황에 대하여 정상인지 여부를 감지하고, 상기 객체 정보에 포함된 적어도 하나의 객체를 분류하는 딥러닝 탐지 모듈; 및 상기 적어도 하나의 객체에 대하여, 사람인지 여부를 판단하고, 상기 상황에 대하여 비정상적이고 상기 적어도 하나의 객체에 사람에 해당하는 객체가 존재하는 경우, 장비 제어부에 긴급 신호를 전송하는 후처리 모듈을 포함하는 것을 특징으로 한다.

기계 학습 모델을 사용하여 편향된 데이터 거절

... 기계 학습 모델(300)을 사용하여 편향된 데이터를 거절하는 방법(500)은, 알려진 비편향 데이터 집단을 포함하는 클러스터 훈련 데이터 세트(130)를 수신하고, 알려진 비편향 데이터 집단의 데이터 특성에 기초하여 수신된 클러스터 훈련 데이터 세트를 클러스터들(212)로 분할하기 위해 클러스터링 모델(211)을 훈련시키는 단계를 포함한다. 클러스터 훈련 데이터 세트의 각 클러스터는 클러스터 가중치(214)를 포함한다. 방법은 또한 기계 학습 모델에 대한 훈련 데이터 세트(302)를 수신하고 클러스터링 모델에 기초하여 기계 학습 모델에 대한 훈련 데이터 세트에 대응하는 훈련 데이터 세트 가중치(218)를 생성하는 단계를 포함한다. 방법은 또한 각각의 클러스터 가중치와 매칭하도록 훈련 데이터 세트 가중치의 각 훈련 데이터 세트 가중치를 조정하고 조정된 훈련 데이터 세트를 비편향 훈련 데이터 세트로서 기계 학습 모델에 제공하는 단계(206)를 포함한다.

Method and device for carrying out eye gaze mapping

The invention relates to a device and a method for performing an eye gaze mapping (M), in which at least one point of vision (B) and/or a viewing direction of at least one person (10) in relation to at least one scene recording (S) of a scene (12) viewed by the at least one person (10) is mapped onto a reference (R). At least a part of an algorithm (A1, A2, A3) for performing the eye gaze mapping (M) is thereby selected from multiple predetermined algorithms (A1, A2, A3) as a function of at least one parameter (P), and the eye gaze mapping (M) is performed on the basis of the at least one part of the algorithm (A1, A2, A3).

Navigation apparatus and method for providing individualization map service thereof

A navigation apparatus and a method for providing an individualization map service of the navigation apparatus are provided. The navigation apparatus includes a detector configured to sense a vehicle state and a travelling environment during travelling, and a processor configured to recognize a drive context based on the vehicle state and the travelling environment, and to make an individualization map to be serviced based on the recognized drive context.

GRAPH CONVOLUTIONAL NETWORKS WITH MOTIF-BASED ATTENTION

Various embodiments describe techniques for making inferences from graph-structured data using graph convolutional networks (GCNs). The GCNs use various pre-defined motifs to filter and select adjacent nodes for graph convolution at individual nodes, rather than merely using edge-defined immediate-neighbor adjacency for information integration at each node. In certain embodiments, the graph convolutional networks use attention mechanisms to select a motif from multiple motifs and select a step size for each respective node in a graph, in order to capture information from the most relevant neighborhood of the respective node.

METHOD OF DETECTING ACTION, ELECTRONIC DEVICE, AND STORAGE MEDIUM

A method of detecting an action, an electronic device, and a storage medium. A method can include: performing a temporal action proposal on at least one target feature data obtained by a feature extraction on a plurality of target frame data of a target resource, so as to obtain at least one first candidate action proposal information; classifying target feature data corresponding to at least one first candidate action proposal interval included in the first candidate action proposal information, so as to obtain at least one classification confidence level corresponding to the at least one first candidate action proposal interval; and determining an action detection result for at least one action segment contained in the target resource according to the at least one classification confidence level corresponding to the at least one first candidate action proposal interval, wherein the action detection result includes an action category and an action period.

Large-scale automated image annotation system

Systems and methods for automating image annotations are provided, such that a large-scale annotated image collection may be efficiently generated for use in machine learning applications. In some aspects, a mobile device may capture image frames, identifying items appearing in the image frames and detect objects in three-dimensional space across those image frames. Cropped images may be created as associated with each item, which may then be correlated to the detected objects. A unique identifier may then be captured that is associated with the detected object, and labels are automatically applied to the cropped images based on data associated with that unique identifier. In some contexts, images of products carried by a retailer may be captured, and item data may be associated with such images based on that retailer's item taxonomy, for later classification of other/future products.

Model training, image processing method, device, storage medium, and program product

The present application provides a model training, image processing method, device, storage medium, and program product relating to deep learning technology, which are able to screen auxiliary image data with image data for learning a target task, and further fuse the target image data and the auxiliary image data, so as to train a built and to-be-trained model with the fusion-processed fused image data. This implementation can increase the amount of data for training the model, and the data for training the model is determined is based on the target image data, which is suitable for learning the target task. Therefore, the solution provided by the present application can train an accurate target model even if the amount of target image data is not sufficient.

METHOD FOR PROVIDING A SINGLING DEVICE OF A STORAGE AND DISPENSING CONTAINER

A system includes a storage and dispensing container for drug portions, a memory and a processor. The process executes instructions to detect by a first detection device rays of light transmitted past a drug portion by a first light emitting source in a first direction, to detect by a second detection device rays of light transmitted past the drug portion by a second light emitting source in a second direction different from the first direction, to generate, by the first and second detection devices, an image of the drug portion, to process the image using image analysis, to determine specified drug information for the drug portion and to identify, based on the determined drug information, a suitable singling device fitting the drug portions to be singularized and to be disposed within the storage and dispensing container. Additional systems for identifying a singling device are also provided.

MACHINE LEARNING BASED OBJECT IDENTIFICATION USING SCALED DIAGRAM AND THREE-DIMENSIONAL MODEL

СПОСОБ И СИСТЕМА ОБЕЗЛИЧИВАНИЯ КОНФИДЕНЦИАЛЬНЫХ ДАННЫХ

Изобретение относится к области защиты данных, а более конкретно к обезличиванию конфиденциальных данных с сохранением структуры данных в текстовых документах. Техническим результатом является обеспечение возможности сохранения стилистической, семантической, лексической и морфологической структуры данных в текстовых документах при их обезличивании, повышение точности обезличивания данных в текстовых документах за счет определения конфиденциальных данных в текстовых документах с помощью модели машинного обучения. Способ обезличивания конфиденциальных данных в текстовых документах с сохранением структуры данных содержит этапы, на которых: получают документ с текстовыми данными; сегментируют и токенизируют текстовые данные; выполняют векторизацию токенов; определяют принадлежности каждого токена к категории конфиденциальных данных с помощью модели машинного обучения; обезличивают данные, относящиеся к токенам с конфиденциальными данными с сохранением структуры данных; формируют список замен ...

СПОСОБ И УСТРОЙСТВО СОПРОВОЖДЕНИЯ МАНЕВРИРУЮЩИХ ЦЕЛЕЙ В ОБЗОРНОЙ ДОПЛЕРОВСКОЙ РЛС

Группа изобретений относится к области радиолокации и может быть использована в обзорных доплеровских радиолокационных станциях при сопровождении траекторий маневрирующих целей. Техническим результатом является повышение устойчивости сопровождения любых типов целей на любых участках их траекторий за счет существенного повышения вероятности корректной привязки к сопровождаемой трассе отметок от цели, движущейся как равномерно прямолинейно, так и интенсивно маневрирующей. В заявленном способе при сопровождении цели в обзорной доплеровской РЛС для коррекции центров стробов по дальности и радиальной скорости относительно их экстраполяции основным трассовым фильтром, а также для вычисления размеров стробов на каждом шаге сопровождения используют результаты автономного сглаживания и экстраполяции измеряемой доплеровской скорости и ускорения, получаемые с помощью специального дополнительного фильтра с полиномиальной моделью первого порядка, который работает параллельно с основным трассовым фильтром ...

СПОСОБ ОПРЕДЕЛЕНИЯ РИСКА РАКА МОЛОЧНОЙ ЖЕЛЕЗЫ НА ОСНОВЕ СПЕКТРОСКОПИИ БИОИМПЕДАНСА И С УЧЕТОМ АНИЗОТРОПИИ БИОМАТЕРИАЛА

Изобретение относится к медицине, а именно к способу определения риска рака молочной железы на основе спектроскопии биоимпеданса и с учетом анизотропии биоматериала. При этом сканируют поверхность кожи молочной железы путем формирования отведений. Для формирования отведений используют четыре матрицы электродов. Отведение образуют подключением биоматериала молочной железы между одним из электродов одной из матриц и всеми электродами одной из смежных матриц, электроды которых объединяют и образуют единый индифферентный электрод. В каждом отведении из графика Коула выделяют шесть дескрипторов из его реальной составляющей и шесть дескрипторов из его мнимой составляющей. Дескрипторы подают на нейронные сети первого иерархического уровня обучаемого классификатора. Решения нейронных сетей первого уровня, принадлежащих одной и той же топологической структуре, агрегируют посредством второго уровня классификатора. Решения нейронных сетей второго иерархического уровня по окончании сканирования подают ...

СПОСОБ ОПРЕДЕЛЕНИЯ БИОЛОГИЧЕСКОГО ВОЗРАСТА ЧЕЛОВЕКА НА ОСНОВЕ БИОХИМИЧЕСКИХ ПОКАЗАТЕЛЕЙ КРОВИ

Изобретение относится к медицине, а именно к профилактической медицине и геронтологии, и может быть использовано для определения биологического возраста человека. Измеряют показатели крови с учетом пола: для мужчин - дегидроэпиандростерон, гликолизированный гемоглобин, соматомедин-С, N-терминальный мозговой натрийуретический пропептид, тестостерон свободный, цистатин-С, С-реактивный белок; для женщин - гомоцистеин, зонулин крови, мочевину, соматомедин-С, цистатин-С, глюкозу, дегидроэпиандростерон. Производят обработку измеренных показателей крови с помощью полносвязной нейронной сети, содержащей пять слоев. Способ обеспечивает высокоточное определение биологического возраста человека за счет использования нейронной сети с функцией обработки биохимических показателей крови. 4 пр.

Deep learning seismic attribute fault predictions

This disclosure presents a fault prediction system using a deep learning neural network, such as a convolutional neural network. The fault prediction system utilizes as input seismic data, and then derives various seismic attributes from the seismic data. In various aspects, the seismic attributes can be normalized and have importance coefficients determined. A sub-set of seismic attributes can be selected to reduce computing resources and processing time. The deep learning neural network can utilize the seismic data and seismic attributes to determine parameterized results representing fault probabilities. The fault prediction system can utilize the fault probabilities to determine fault predictions which can be represented as a predicted new seismic data, such as using a three-dimensional image.

Determination of health status of vehicular systems in vehicles

Determination of Health Status of Vehicular Systems In Vehicles A predictive maintenance method for determining a health status of a vehicular system included in a first vehicle is provided. A first dataset is collected for a plurality of vehicles. The first dataset is processed to determine a plurality of features corresponding to the vehicular system and a plurality of feature values for the plurality of features. A classifier is trained, based on the plurality of features, for determining the health status of the vehicular system. A second dataset including second operational data, second vehicle data, second trip data, and second service data for the first vehicle is collected. The second dataset is provided as input to the trained classifier. The health status of the vehicular system included in the first vehicle is determined based on an output of the trained classifier for the second dataset.

Training object detection systems with generated images

Apparatuses, systems, and techniques to identify objects within an image using self-supervised machine learning. In at least one embodiment, a machine learning system is trained to recognize objects by training a first network to recognize objects within images that are generated by a second network which may be a controllable synthesis network. The one or more neural networks may be controlled by an input that specifies a location of an object to add to an image with the image being used to train the object detection neural network. The one or more neural networks may include a first network that generates a representation of an object and a second network that generates a background image and a combination of the background image and the representation of the object may be used to train the object detection neural network.

Systems and methods for generating a medical image

A method for generating a composite 2D interpretation image, comprising inputting each 2D medical image of a divided 3D image, into a 2D classifier (106,306) trained on 2D images labelled with an indication of a visual finding, computing a respective explanation map for each respective 2D image(108,308), the respective explanation map including regions corresponding to corresponding regions of the respective 2D image, each respective region of the respective explanation map is associated with a computed explainable weight indicative of an influence of the respective 2D image, and computing a composite 2D interpretation image(114,310) comprising a respective aggregation weight for each respective region thereof, each respective aggregation weight computed by aggregating the explainable weights computed for the respective regions corresponding to the respective region of the composite 2D interpretation image.

Deep learning seismic attribute fault predictions

행동별 이상 감지 시스템

... 본 발명은 객체의 행동에 따라 신뢰도를 높일 수 있는 판단 요소를 활용하여 객체 또는 군집의 이상 상태를 판단할 수 있는 행동별 이상 감지 시스템을 제공하는데 그 목적이 있다. 상기한 목적을 이루기 위해 본 발명에 따른 행동별 이상 감지 시스템은 객체 또는 군집을 촬영하여 이미지 정보를 생성하는 카메라부 및 상기 이미지 정보를 수신하여 상기 객체 또는 상기 군집의 행동을 분류하여 행동 분류 정보를 생성하는 분류부, 상기 행동 분류 정보에 기초하여 분류된 행동에 따른 판단 데이터를 생성하는 데이터 생성부 및 상기 판단 데이터를 학습하여 상기 이미지 정보를 기초로 분류된 행동에 따른 상기 객체 또는 상기 군집의 이상 여부를 판단하는 판단부를 포함하는 정보 처리부를 포함하여 이루어지는 것을 특징으로 할 수 있다.

림프종 판단 시스템 및 방법

... 본 명세서는 복수의 타깃 세포들에 대한 Ki-67 염색 패턴 이미지를 입력받는 입력부, Ki-67 염색 패턴 이미지의 염색 강도를 분석하여, 복수의 타깃 세포들을 복수의 그룹들으로 분류하는 이미지 분석부, 복수의 그룹들 각각에 포함된 타깃 세포들의 개수를 기초로, 복수의 타깃 세포들에 대한 림프종 타입이 미만성 거대 B세포 림프종(DLBCL, Diffuse large B-cell lymphoma)인지 또는 버킷 림프종(BL, Burkitt lymphoma)인지를 판단하는 판단부를 포함하는 림프종 판단 시스템 및 그에 대한 방법을 제공한다.

인공지능 기반의 병해충 진단 시스템 및 방법

... 본 발명은 인공지능 기반의 병해충 진단 시스템 및 방법에 관한 것으로서, 미리 수집된 원시 작물 이미지를 저장하고 있는 이미지 저장 모듈; 이미지 저장 모듈에 저장된 원시 작물 이미지에 비지도 학습(Unsupervised Learning)에 따른 Auto ML(Auto Machine Learning)을 적용하는 방식으로 원시 작물 이미지를 그 클래스별로 분류하여 분류 작물 이미지를 생성하는 Auto ML 모듈; Auto ML 모듈에 의해 생성된 분류 작물 이미지에 미리 정의된 Auto Labeling 알고리즘을 적용하여 분류 작물 이미지의 Annotation 파일을 생성하는 Auto Labeling 모듈; Auto ML 모듈에 의해 생성된 분류 작물 이미지와 Auto Labeling 모듈에 의해 생성된 Annotation 파일을 매핑하여 학습 데이터 셋(Training Data Set)을 생성하는 학습 데이터 생성 모듈; 학습 데이터 생성 모듈에 의해 생성된 학습 데이터 셋을 기반으로, 입력된 작물 이미지 상의 병해충을 신경망을 통해 진단하도록 구성된 딥 러닝(Deep Learning) 모델을 학습시키는 학습 모듈; 및 제1 사용자가 소지한 제1 사용자 단말로부터 대상 작물 이미지를 입력받으며, 학습 모듈에 의해 학습된 딥 러닝 모델에 대상 작물 이미지가 적용되어 도출된, 대상 작물 이미지에 반영된 병해충 진단 결과를 제1 사용자 단말로 제공하는 인터페이스 모듈;을 포함하는 것을 특징으로 한다.

미입력 라벨 검출 시스템 및 방법

... 본 발명은 미입력 라벨 검출 시스템 및 방법에 관한 것이다. 본 발명에 따른 미입력 라벨 검출 시스템은 영상을 입력받는 영상 입력부, 특이 데이터 여부를 판단하기 위한 이상 검출 네트워크 추론부 및 상기 입력된 영상에 대한 추론 영상을 생성하기 위한 의미 분할 네트워크 추론부를 포함하고, 상기 이상 검출 네트워크 추론부는 상기 입력된 영상이 특이 데이터라고 판단된 경우, 신뢰도 기준을 추가하여 신뢰도 기준이 추가된 상기 영상을 상기 의미 분할 네트워크 추론부로부터 전달 받은 객체 크기 계산 값과 취합하여 객체 크기 별 요구 신뢰도를 계산한다.

APPARATUS AND METHOD FOR PROVIDING MICROORGANISM INFORMATION

... 본 발명의 일 실시예는, 시료로부터 출사되는 출사파동을 시계열 순으로 촬영한 복수의 영상을 수신하는 수신부, 상기 시계열 순으로 촬영한 복수의 영상으로부터 시간에 따른 변화의 특징(feature)을 추출하는 검출부, 상기 추출된 특징을 기초로 분류기준을 기계학습하는 학습부 및 상기 분류기준을 기초로 시료에 포함된 미생물의 종류 또는 농도를 구분하는 판단부를 포함하고, 상기 복수의 영상 각각은 상기 시료로 입사되는 파동에 기인하여 상기 미생물에 의해 다중산란(multiple scattering)되어 발생되는 스펙클(speckle) 정보를 포함하는, 미생물 정보 제공 장치를 제공한다.

Automatic update scheduler systems and processes

A computer-implemented method includes: registering, by a computer device, a device to a network; collecting, by the computer device, device data from the device through the network; compiling, by the computer device, training data from the collected device data; training, by the computer device, a machine learning model using the training data; predicting, by the computer device and using the machine learning model, a time when the device will be in an inactive system state; and automatically scheduling, by the computer device and based on the predicting, an application of an update for the time when the device is in the inactive system state.

IMAGING APPARATUS

An imaging apparatus includes imaging units that image a subject, a title generation executing unit that generates title images each including a display title for classifying a plurality of captured images acquired by the imaging unit and inserts the title image into the captured images, a storage unit that stores the captured images and the title images, a display unit that displays the captured images and the title images stored in the storage unit in the form of a list, and an image search executing unit that searches for a desired captured image from the plurality of captured images and the title images displayed on the display unit in the form of the list. The title generation executing unit hierarchically generates the title images in accordance with a priority set in each type of display title for classifying the captured images.

Training a machine to recognize a motor vehicle driver using a mobile device

Determining that a motor vehicle driver is using a mobile device while driving a motor vehicle. Multiple images of a driver of a motor vehicle are captured through a side window of the motor vehicle. Positive images show a driver using a mobile device while driving a motor vehicle. Negative images show a driver not using a mobile device while driving a motor vehicle. Multiple training images are selected from both the positive images and the negative images. The selected training images and respective labels, indicating that the selected training images are positive images or negative images, are input to a machine (e.g. Convolutional Neural Network, (CNN)). The CNN is trained to classify that a test image, captured through a side window of a motor vehicle, shows a driver using a mobile device while driving the motor vehicle.

System and method for dynamically configurable deep brain stimulation

A DBS system and method for predicting future neurological activity in a subject and administering a corrective electrical stimulation signal to prevent anticipated pathological neuronal activity. The DBS system includes an implantable electrode configured to both record neuronal activity from a target brain area in a subject and administer the corrective electric stimulation signal to the target area. The DBS system also includes a controller configured to determine the characteristics of the corrective electrical stimulation signal based on point process models of healthy and pathological neuronal activity in the target area.

Method for a Pattern Discovery and Recognition

A method is for a pattern discovery and recognition, wherein a first sequence comprising first sequence symbols relating to a concept and a tag associated to the first sequence are received, transition probability matrices are obtained from transition frequency matrices representing the frequency data of the occurrences of the transitions between the first sequence symbols at different distances in the first sequence, and the transition probability matrices for each tag and each distance are learnt for obtaining an activation function determining the concept occurring in a second sequence. A computer program product and an apparatus are for executing the pattern discovery and recognition method.

Detecting counterfeit electronic components using emi telemetric fingerprints

One embodiment of the present invention provides a system that non-intrusively detects counterfeit components in a target computer system. During operation, the system collects target electromagnetic interference (EMI) signals generated by the target computer system using one or more antennas positioned in close proximity to the target computer system. The system then generates a target EMI fingerprint for the target computer system from the target EMI signals. Next, the system compares the target EMI fingerprint against a reference EMI fingerprint to determine whether the target computer system contains a counterfeit component.

Computer-readable medium storing learning-model generating program, computer-readable medium storing image-identification-information adding program, learning-model generating apparatus, image-identification-information adding apparatus, and image-identification-information adding method

A computer-readable medium storing a learning-model generating program causing a computer to execute a process is provided. The process includes: extracting feature values from an image for learning that is an image whose identification information items are already known, the identification information items representing the content of the image; generating learning models by using binary classifiers, the learning models being models for classifying the feature values and associating the identification information items and the feature values with each other; and optimizing the learning models for each of the identification information items by using a formula to obtain conditional probabilities, the formula being approximated with a sigmoid function, and optimizing parameters of the sigmoid function so that the estimation accuracy of the identification information items is increased.

Method and System For Patient-Specific Modeling of Blood Flow

Embodiments include a system for planning treatment for a patient. The system may include at least one computer system configured to receive patient-specific data regarding a geometry of an anatomical structure of the patient, create a three-dimensional model representing at least a portion of the anatomical structure of the patient based on the patient-specific data, and determine a first fractional flow reserve within the anatomical structure of the patient based on the three-dimensional model and information regarding a physiological condition of the patient. The at least one computer system may be further configured to receive input from a user regarding a plan of treatment, modify the physiological condition of the patient based on the received input, and determine a second fractional flow reserve within the anatomical structure of the patient based on the modified physiological condition of the patient.

Storing and Searching Fingerprints Derived from Media Content Based on a Classification of the Media Content

Attributes are identified in media content. A classification value of the media content is computed based on the identified attributes. Thereafter, a fingerprint derived from the media content is stored or searched for based on the classification value of the media content.

Digital image analysis using multi-step analysis

Systems and methods for implementing a multi-step image recognition framework for classifying digital images are provided. The provided multi-step image recognition framework utilizes a gradual approach to model training and image classification tasks requiring multi-dimensional ground truths. A first step of the multi-step image recognition framework differentiates a first image region from a remainder image region. Each subsequent step operates on a remainder image region from the previous step. The provided multi-step image recognition framework permits model training and image classification tasks to be performed more accurately and in a less resource intensive fashion than conventional single-step image recognition frameworks.

Method and system for determining whether a drug will be effective on a patient with a disease

A process of determining whether a patient with a disease or disorder will be responsive to a drug, used to treat the disease or disorder, including obtaining a test spectrum produced by a mass spectrometer from a serum produced from the patient. The test spectrum may be processed to determine a relation to a group of class labeled spectra produced from respective serum from other patients having the or similar clinical stage same disease or disorder and known to have responded or not responded to the drug. Based on the relation of the test spectrum to the group of class labeled spectra, a determination may be made as to whether the patient will be responsive to the drug.

System and Method for Classifying Data Streams with Very Large Cardinality

An object and attributes that describe that object are identified. The attributes are grouped into attribute patterns, and classification classes are identified. For each identified class a sketch table containing a plurality of parallel hash tables is created. For the object to be classified, each attribute pattern is processed using the all of the hash functions for each sketch table, resulting in a plurality of values under each sketch table for a single attribute pattern. The lowest value is selected for each sketch table. The distribution of values across all sketch tables is evaluated for each attribute pattern, producing a discriminatory power for each attribute pattern. Attribute patterns having a discriminatory power above a given threshold are selected and added to the associated sketch table values. The sketch table with the largest overall sum is identified, and the associated class is assigned to the object belonging to the attribute patterns.

Method and apparatus for machine error detection by combining multiple sensor inputs

One of the most significant safety concerns in the automation of extracorporeal blood treatments such as dialysis is the risk of blood leakage. Extracorporeal blood treatment systems draw blood at such a high rate that a loss of integrity in the blood circuit can be serious. There are a number of mechanisms for detecting and preventing leaks, but none is perfect. This tends to limit the use of such equipment in unsupervised settings, such as the home will be limited. Some leak detection schemes can be made sensitive enough to detect the barest of leaks, but when this is done, they result in too many false positives. The invention combines information from multiple inputs to enhance sensitivity in leak detection and reduce the problem of false positives.

Method for Implementing a High-Level Image Representation for Image Analysis

Robust low-level image features have been proven to be effective representations for a variety of visual recognition tasks such as object recognition and scene classification; but pixels, or even local image patches, carry little semantic meanings. For high-level visual tasks, such low-level image representations are potentially not enough. The present invention provides a high-level image representation where an image is represented as a scale-invariant response map of a large number of pre-trained generic object detectors, blind to the testing dataset or visual task. Leveraging on this representation, superior performances on high-level visual recognition tasks are achieved with relatively classifiers such as logistic regression and linear SVM classifiers.

System and method for effectively performing a scene rectification procedure

A system and method for effectively performing a scene rectification procedure comprises an image manager that includes a segmentation module, a label module, and a rectification module. The segmentation module initially performs a segmentation procedure upon an image to produce corresponding sub-scenes. The label module then categorizes the sub-scenes by assigning initial labels without utilizing context information from other sub-scenes in the image. The rectification module performs a semantic grouping procedure upon the sub-scenes to produce semantic group nodes corresponding to pairs of the sub-scenes that have a predefined semantic relationship. The rectification module converts a sub-scene graph of the sub-scenes into a semantic graph that includes the semantic group nodes. The rectification module then performs a rectification procedure to convert the initial labels of the sub-scenes into rectified labels. A processor of an electronic device typically controls the image manager for performing the scene rectification procedure.

Robust feature matching for visual search

Techniques are disclosed for performing robust feature matching for visual search. An apparatus comprising an interface and a feature matching unit may implement these techniques. The interface receives a query feature descriptor. The feature matching unit then computes a distance between a query feature descriptor and reference feature descriptors and determines a first group of the computed distances and a second group of the computed distances in accordance with a clustering algorithm, where this second group of computed distances comprises two or more of the computed distances. The feature matching unit then determines whether the query feature descriptor matches one of the reference feature descriptors associated with a smallest one of the computed distances based on the determined first group and second group of the computed distances.

System and method for disaggregating power load

Systems and methods of disaggregating power load are provided. An example of a method is carried out by program code stored on non-transient computer-readable medium and executed by a processor. The method includes receiving time series data representing total energy consumption. The method also includes identifying distinguishing features in the time series data. The method also includes identifying energy consumption constituents of the total energy consumption based on the features.

Method and apparatus to detect coronary artery calcification or disease

Coronary artery calcification (CAC) occurs at an earlier age in diabetes and is a risk factor for coronary artery disease (CAD) in subjects with or without diabetes. One postulated mechanism for the increased CAC is the accelerated accumulation of advanced glycation end products (AGEs) in the vasculature. As certain collagen AGEs fluoresce, skin intrinsic fluorescence (SIF) can act as a novel maker of collagen AGEs levels. The present invention provides methods and apparatuses for detecting SIF that can be a useful marker of CAD risk and a therapeutic target.

Information processing apparatus, information processing method, and program

A probability function with highest likelihood is calculated based on data. A canonical distribution in statistical physics and a temperature parameter of the canonical distribution are calculated as a fluctuation of the data. A probability function is estimated using the calculated probability function with the highest likelihood, the calculated fluctuation, and the canonical distribution. The present technology is applicable to an apparatus that estimates and uses a probability function.

Classification method and apparatus

A method and system for building a classification model for classifying documents comprising: representing each of a plurality of documents by a vector of n dimensions, said n dimensions forming a vector space; and representing the classification of already classified documents into classes by separating said vector space into a plurality of subspaces by one or more hyperplanes.

Method For Calculating Weight Ratio By Quality Grade In Grain Appearance Quality Grade Discrimination Device

A method is provided for calculating a weight ratio by quality grade using a grain appearance quality grade discrimination device. The method involves the steps of imaging a plurality of grains; discriminating the quality grade of the grains on the basis of data of the imaged grains; tallying, by quality grade, the number of pixels in said data of the imaged grains with regards to the grains whose quality grade has been discriminated; multiplying the number of pixels tallied by quality grade by a weight conversion coefficient per pixel predetermined by quality grade, and thereby converting said number of pixels into a weight by quality grade; and calculating the weight ratio by quality grade of the grains on the basis of the weight by quality grade.

Method and system for patient-specific modeling of blood flow

Embodiments include a system for determining cardiovascular information for a patient. The system may include at least one computer system configured to receive patient-specific data regarding a geometry of the patient's heart, and create a three-dimensional model representing at least a portion of the patient's heart based on the patient-specific data. The at least one computer system may be further configured to create a physics-based model relating to a blood flow characteristic of the patient's heart and determine a fractional flow reserve within the patient's heart based on the three-dimensional model and the physics-based model.

Method and system for patient-specific modeling of blood flow

Embodiments include a system for determining cardiovascular information for a patient. The system may include at least one computer system configured to receive patient-specific data regarding a geometry of the patient's heart, and create a three-dimensional model representing at least a portion of the patient's heart based on the patient-specific data. The at least one computer system may be further configured to create a physics-based model relating to a blood flow characteristic of the patient's heart and determine a fractional flow reserve within the patient's heart based on the three-dimensional model and the physics-based model.

Method for determining a preferred node in a classification and regression tree for use in a predictive analysis

Techniques are described for determining what node of a classification and regression tree (CART) should be used by a predictive analysis application. A first approach is to use a standard deviation of the data at a given the level of the CART to determine whether data in the next, lower node is more consistent than the data in the current node. A second approach is to measure a correlation between data points in a given node and the time at which each point was sampled (or other correlation metric) to identify a preferred node.

Depth Image Compression

Depth image compression is described for example, to enable body-part centers of players of a game to be detected in real time from depth images or for other applications such as augmented reality, and human-computer interaction. In an embodiment, depth images which have associated body-part probabilities, are compressed using probability mass which is related to the depth of an image element and a probability of a body part for the image element. In various examples, compression of the depth images using probability mass enables body part center detection, by clustering output elements, to be speeded up. In some examples, the scale of the compression is selected according to a depth of a foreground region and in some cases different scales are used for different image regions. In some examples, certainties of the body-part centers are calculated using probability masses of clustered image elements.

Short-term Load Forecast Using Support Vector Regression and Feature Learning

In a support vector regression approach to forecasting power load in an electrical grid, a feature learning scheme weights each feature in the input data with its correlation with the predicted load, increasing the prediction accuracy. The kernel matrix for the input training data is computed such that features that align better with the target variable are given greater weight. The resulting load forecast may be used to compute commands sent to demand response modules.

Methods and Apparatus for Chatter Reduction in Video Object Segmentation Using a Variable Bandwidth Search Region

Systems, methods, and computer-readable storage media for chatter reduction in video object segmentation using a variable bandwidth search region. A variable bandwidth search region generation method may be applied to a uniform search region to generate a variable bandwidth search region that reduces the search range for segmentation methods such as a graph cut method. The method may identify parts of the contour that are moving slowly, and reduce the search region bandwidth in those places to stabilize the segmentation. This method may determine a bandwidth for each of a plurality of local windows of an image according to an estimate of how much an object in the image has moved from a previous image. The method may blend the bandwidths for the plurality of local windows to generate a blended map. The method may then generate a variable bandwidth search region for an object according to the blended map.

Information processing method, information processing apparatus, and recording medium

The robustness of discriminating results at each stage is improved in discrimination processing in which a plurality of stages of discriminators are used to identify an object. An information processing apparatus in which a plurality of stages of the discriminators are used to identify a class of an object, comprises a candidate class output unit that acquires as a candidate class a class discriminated at a first stage of the discriminators, and an extended class setting unit that sets in a second stage of the discriminators, a class of a second stage of the discriminators, which is defined as an extended partial space of a partial space defined by a candidate class in a discriminating space used in discriminating the candidate class by the first stage of the discriminators, as a class to be discriminated at this second stage of the discriminators.

Method, system and apparatus for determining a subject and a distractor in an image

A method of identifying a subject and a distractor in a target image is disclosed. The method receives a reference image comprising image content corresponding to image content of the target image. A first saliency map, which defines a distribution of visual attraction values identifying salient regions within the target image, and a second saliency map, which defines a distribution of visual attraction values identifying salient regions within the reference image, are determined. The method compares image content in salient regions of the first saliency map and the second saliency map. The subject is identified by a salient region of the target image sharing image content with a salient region of the reference image. The distractor is identified based on at least one remaining salient region of the target image.

Image Classification

The present disclosure introduces a method and an apparatus for classifying images. Classification image features of an image for classification are extracted. Based on a similarity relationship between each classification image feature and one or more visual words in a pre-generated visual dictionary, each classification image feature is quantified by multiple visual words in the visual dictionary and a similarity coefficient between each classification image feature and each of the visual words is determined. Based on the similarity coefficient of each visual word that corresponds to different classification image features, a weight of each visual word is determined to establish a classification visual word histogram of the image for classification. The classification visual word histogram is input into an image classifier that is trained by sample visual word histograms arising from multiple sample images. An output result is used to determine a classification of the image for classification.

Prediction of horizontally transferred gene

Method(s) and system(s) for identifying horizontally transferred genes are described herein. The method includes defining a cuboid in a three dimensional space, wherein the cuboid includes fragment points corresponding to the genomic fragments belonging to a plurality of sequenced microbial genomes, and dividing the cuboid into a plurality of grids. The method further includes selecting one or more grids corresponding to a selected genome and classifying each of the selected grids as one of majority, minority, and mixed grids, based on number of fragment points corresponding to the selected genome in each of the selected grids. Further, at least one genomic fragment from the minority and the mixed grids is identified as the horizontally transferred gene based on a distance ratio assessment.

Method and apparatus for multi-label segmentation

A method and an apparatus for multi-label segmentation of an image are described. First an energy function is determined for the image. Then for a homogeneous region of the image variables of the energy function are grouped to a single variable. Subsequently the energy function is minimized and labels are assigned to regions of the image based on the minimized energy function.

Dynamic phase machine automation of oil & gas processes

A method, apparatus, and program product facilitate the automation of an oil & gas process, e.g., a drilling process, through the use of a dynamic phase machine incorporating multiple autonomous agents.

Image classification apparatus and recording medium having program recorded therein

An image-classification apparatus includes: a first feature extraction unit to acquire a feature value of each of block images obtained by segmenting an input image; an area-segmentation unit to assign each of the block images to any one of K areas based on the feature value; a second feature extraction unit to acquire, based on an area-segmentation result, a feature vector whose elements including, the number of adjacent spots, each including two block images adjacent to each other in the input image, for each combination of the areas whereto the two block images are assigned; or a ratio of the number of block images assigned to each of the K areas to all the number of block images adjacent to a block image assigned to each of the K areas; and a classification unit to classify to which of a plurality of categories the input image belong.

System and method for root cause analysis of mobile network performance problems

A method and system for identifying causes of performance metric changes in a network by selecting, from a pool of network event counters, a plurality of candidate counters relevant to a performance metric; grouping the candidate counters into clusters of similar counters; selecting, from each cluster, one or more representative counters; and fitting the selected representative counters to a model of the performance metric to determine thereby a set of representative counters most relevant to the performance metric.

Methods and systems for computer-based selection of identifying input for class differentiation

In systems and methods for computer-based selection of identifying input for differentiating classes, training regions (each of which is associated with a defined class) are specified in a training space that is organized by data bands according to selected definitions. Windows are defined in training elements associated with data locations in the training regions. Multiple training windows are defined in the training elements in a known band in the training data. Relevance measures for training windows represent an extent of likelihood of correctly identifying class for a test location based on data band, window position within the training element, and the frequency of occurrence of data symbols in training windows at the window position. The window having the highest value relevance measure is selected as the most relevant window. Multiple most relevant windows, together with their parameters, are selected as identifying input to facilitate class differentiation in test spaces.

Image Processing Apparatus that Groups Objects Within Image

Provided is an image processing apparatus including: a grouping preference unit configured to register user preference information on a storage device based on a user operation, the user preference information indicating how objects within an image are to be classified into groups; an image analysis unit configured to detect the objects within the image; and a grouping unit configured to read the user preference information from the storage device and classify the objects detected within the image into the groups indicated in the read user preference information.

Identification of duplicates within an image space

Implementations for identifying duplicate images in an image space are described. An image space is partitioned into a plurality of coarse clusters based on signatures of the images within the image space. The signatures are determined from compact descriptors of the images. Refined clusters that include one or more images of an individual coarse cluster are created based on pair-wise comparisons of the compact descriptors of images in the coarse cluster, and the refined clusters are identified as sets of duplicate images. The refined clusters are grown by searching in similar coarse clusters for images to add to the refined clusters.

Methods and systems for employing artificial intelligence in automated orthodontic diagnosis & treatment planning

Methods and systems for diagnosing and identifying a treatment for an orthodontic condition are disclosed. Such methods and systems generally entail the use of a server on which a centralized website is hosted. The server is configured to receive patient data through the website. The methods and systems further include the use of a database that includes or has access to (i) information derived from textbooks and scientific literature and (ii) dynamic results derived from ongoing and completed patient treatments. The methods and systems further include the operation of at least one computer program within the server, which is capable of analyzing the patient data and identifying at least one diagnosis of the orthodontic condition. The methods and systems further entail assigning a probability value to the at least one diagnosis, with the probability value representing a likelihood that the diagnosis is accurate. The methods and systems further include instructing the computer program to identify at least one treatment approach, a corrective appliance, or a combination thereof for the at least one diagnosis.

Systems and methods for extraction of policy information

In a system for extracting policy information from text, a processor analyzes if the text is relevant to a top-level category, and then determines if at least a portion of the text is relevant to categories and subcategories within a taxonomy of categories and subcategories related to the top-level category. If at least a portion of the text is determined to be relevant to the category/subcategory, a classifier extracts policy information associated with the category/subcategory. Using text that includes a known policy the classifiers can be trained to correctly recognize categories/subcategories, and the values associated therewith.

Validation template for valuable media of multiple classes

A technique for use in automatic validation of a media item involves accessing a template that comprises multiple one-class classifiers, each corresponding to one of multiple classes to which the media item might belong, and then applying each of the one-class classifiers to an image of the media item to generate a result set for each of the multiple classes. The result set for each media class is then analyzed to assess whether the media item belongs to that class.

Comparison method and comparator for graphic comparison of targets

A method, a data network arrangement and a computer program product by which a comparator can be implemented in a data network. The user of the comparator stores his evaluation of the properties included in the targets to be evaluated in a database in the server in the data network using a graphic one- or two-dimensional evaluation frame. In the comparator the comparator user's evaluations are compared to property evaluations given by a reference user. When calculating the total accuracy percentage of the comparison the evaluations given by the comparator user are weighted with a weighting coefficient, which is obtained by normalising first the importance evaluations of all the properties given by the reference user and by using the normalised importance evaluation of a certain property as the weighting coefficient for this property.

Stationary target detection by exploiting changes in background model

A computer-implemented method for processing one or more video frames may include obtaining one or more video frames; generating one or more blobs using the one or more video frames; classifying the one or more blobs to produce one or more classified blobs, wherein the one or more classified blobs include one or more of a stationary target, a moving target, a target insertion, a target removal, or a local change; and constructing a list of detected targets based on the one or more classified blobs.

Methods and apparatus for performing transformation techniques for data clustering and/or classification

Some aspects include transforming data for which at least one constraint has been specified on a portion of the data, the at least one constraint relating to a similarity and/or dissimilarity of at least some of the portion of the data. Techniques comprise determining a first transformation that approximates the at least one constraint using a cosine similarity as a measure of the similarity and/or dissimilarity of the at least a portion of the data, applying at least the first transformation to the data to obtain transformed data, and fitting a plurality of clusters to the transformed data to obtain a plurality of established clusters. Some aspects include classifying input data by transforming the input data using at least the first transformation and comparing the transformed input data to the established clusters.

MEMS Fast Focus Camera Module

An auto focus camera module includes a camera module housing defining an aperture and an internal cavity to accommodate camera module components, an image sensor coupled to or within the housing, a lens barrel within the housing that contains an optical train including at least one movable lens disposed relative to the aperture and image sensor to focus images of scenes onto the image sensor along an optical path, and a fast focus MEMS actuator coupled to one or more lenses of the optical train including the at least one movable lens and configured to rapidly move said at least one movable lens relative to the image sensor to provide autofocus for the camera module in each frame of a preview or video sequence or both.

Video segmentation method

A system and method implemented as a software tool for foreground segmentation of video sequences in real-time, which uses two Competing 1-class Support Vector Machines (C-1SVMs) operating to separately identify background and foreground. A globalized, weighted optimizer may resolve unknown or boundary conditions following convergence of the C-1SVMs. The objective of foreground segmentation is to extract the desired foreground object from live input videos, with fuzzy boundaries captured by freely moving cameras. The present disclosure proposes the method of training and maintaining two competing classifiers, based on Competing 1-class Support Vector Machines (C-1SVMs), at each pixel location, which model local color distributions for foreground and background, respectively. By introducing novel acceleration techniques and exploiting the parallel structure of the algorithm (including reweighing and max-pooling of data), real-time processing speed is achieved for VGA-sized videos.

Recording media processing device, control method of a recording media processing device, and storage medium

The recognition rate is improved and recognition errors are suppressed when recognizing magnetic ink characters. A character recognition unit 80 calculates the difference between the reference waveform data of each character in a character set and the character waveform data of a read magnetic ink character 101 , and defines the characters with the smallest differences to the read character as first and second candidate characters. If scaling the reference waveforms of the first and second candidate characters creates waveforms that are similar with a smaller difference therebetween than before scaling, and the ratio between the difference B between the waveform of the second candidate and the read character, and the difference A between the waveform of the first candidate and the read character, is greater than or equal to a specific value, the character recognition unit scales and adjusts the reference waveforms to recognize the magnetic ink character 101.

Real time hand tracking, pose classification, and interface control

A hand gesture from a camera input is detected using an image processing module of a consumer electronics device. The detected hand gesture is identified from a vocabulary of hand gestures. The electronics device is controlled in response to the identified hand gesture. This abstract is not to be considered limiting, since other embodiments may deviate from the features described in this abstract.

System and method for biometric authentication in connection with camera equipped devices

The present invention relates generally to the use of biometric technology for authentication and identification, and more particularly to non-contact based solutions for authenticating and identifying users, via computers, such as mobile devices, to selectively permit or deny access to various resources. In the present invention authentication and/or identification is performed using an image or a set of images of an individual's palm through a process involving the following key steps: (1) detecting the palm area using local classifiers; (2) extracting features from the region(s) of interest; and (3) computing the matching score against user models stored in a database, which can be augmented dynamically through a learning process.

Dynamic bayesian networks for vehicle classification in video

A system and method for classification of passenger vehicles and measuring their properties, and more particularly to a stochastic multi-class vehicle classification system, which classifies a vehicle (given its direct rear-side view) into one of four classes Sedan, Pickup truck, SUV/Minivan, and unknown, and wherein a feature pool of tail light and vehicle dimensions is extracted which feeds a feature selection algorithm to define a low-dimensional feature vector, and the feature vector is then processed by a Hybrid Dynamic Bayesian Network (HDBN) to classify each vehicle.

Sentiment classification using out of domain data

Providing sentiment classification of out of domain data are disclosed herein. In some aspects, a source domain having a trained classifier is matched to a target domain having a target classifier. The trained classifier may include identifiers that may be used to predict the sentiment of opinion data for the source domain. The target classifier may use the identifiers of the trained classifier to determine the sentiment of opinion data for the target domain.

Neural network with lane aggregation for lane selection prediction of moving objects during autonomous driving

In one embodiment, an autonomous driving system of an ADV perceives a driving environment surrounding the ADV based on sensor data obtained from various sensors, including detecting one or more lanes and at least a moving obstacle or moving object. For each of the lanes identified, an NN lane feature encoder is applied to the lane information of the lane to extract a set of lane features. For a given moving obstacle, an NN obstacle feature encoder is applied to the obstacle information of the obstacle to extract a set of obstacle features. Thereafter, a lane selection predictive model is applied to the lane features of each lane and the obstacle features of the moving obstacle to predict which of the lanes the moving obstacle intends to select.

Autonomous driving method and apparatus

The present disclosure provides an autonomous driving method and an apparatus. The method includes: receiving a currently collected image transmitted by a unmanned vehicle, where the currently collected image is an image collected in a target scenario; acquiring current driving data according to the currently collected image and a pre-trained autonomous driving model, where the autonomous driving model is used to indicate a relationship between an image and driving data in at least two scenarios, and the at least two scenarios include the target scenario; and sending the current driving data to the unmanned vehicle. Robustness of the unmanned driving method is improved.

Method and appartaus for data efficient semantic segmentation

A method and system for training a neural network are provided. The method includes receiving an input image, selecting at least one data augmentation method from a pool of data augmentation methods, generating an augmented image by applying the selected at least one data augmentation method to the input image, and generating a mixed image from the input image and the augmented image.

Hierarchical classification in credit card data extraction

Embodiments herein provide computer-implemented techniques for allowing a user computing device to extract financial card information using optical character recognition (“OCR”). Extracting financial card information may be improved by applying various classifiers and other transformations to the image data. For example, applying a linear classifier to the image to determine digit locations before applying the OCR algorithm allows the user computing device to use less processing capacity to extract accurate card data. The OCR application may train a classifier to use the wear patterns of a card to improve OCR algorithm performance. The OCR application may apply a linear classifier and then a nonlinear classifier to improve the performance and the accuracy of the OCR algorithm. The OCR application uses the known digit patterns used by typical credit and debit cards to improve the accuracy of the OCR algorithm.

Classification method and system for high-throughput transparent articles

A method for detecting and classifying defects in high-throughput transparent articles such as syringes, vials, cartridges, ampules, and bottles is provided. The method includes the steps of providing a stream of the articles; capturing a first digital image of each of the articles in the stream; inspecting the first digital image for objects; determining parameters of the objects; performing a first classification step to classify the objects into a first defect class and a second defect class; performing a second classification step to classify the objects into a plurality of defect types using at least two second classification models; comparing at least one object parameter of a classified object with a predetermined defect type dependent threshold; classifying the article as defective or non-defective based on the comparing step; and separating defective articles from non-defective articles.

Method and apparatus for enhancing semantic features of sar image oriented small set of samples

The present disclosure relates to a method for enhancing sematic features of SAR image oriented small set of samples, comprising: acquiring a sample set of an SAR target image, and performing transfer learning and training on the sample set to obtain a initialized deep neural network of an SAR target image, the sample set comprising an SAR target image and an SAR target virtual image; performing network optimization on the deep neural network by an activation function, and extracting features of the SAR target image by the optimized deep neural network to obtain a feature map; and mapping, by an auto-encoder, the feature map between a feature space and a semantic space to obtain a deep visual feature with an enhanced semantic feature.

System and Method for Control Constrained Operation of Machine with Partially Unmodeled Dynamics Using Lipschitz Constant

A control system for controlling a machine with partially modeled dynamics to perform a task estimates a Lipschitz constant bounding the unmodeled dynamics of the machine, initializes a constraint-admissible control policy using the Lipschitz constant for controlling the machine to perform a task, such that the constraint-admissible control policy satisfies stability constraint, safety and admissibility constraint including one or combination of a state constraint and an input constraint, and has a finite cost on the performance of the task, and jointly controls the machine and update the control policy to control an operation of the machine to perform the task according the control policy starting with the initialized constraint-admissible control policy and to update the control policy using data collected while performing the task. In such a manner, the updated control policy is constraint-admissible.

System and method for automatic pulmonary embolism detection

A system and method for detecting pulmonary embolisms in a subject's vasculature are provided. In some aspects, the method includes acquiring a set of images representing a vasculature of the subject, and analyzing the set of images to identify pulmonary embolism candidates associated with the vasculature. The method also includes generating, for identified pulmonary embolism candidates, image patches based on a vessel-aligned image representation, and applying a set of convolutional neural networks to the generated image patches to identify pulmonary embolisms. The method further includes generating a report indicating identified pulmonary embolisms.

High resolution 3d point clouds generation from downsampled low resolution lidar 3d point clouds and camera images

In one embodiment, a method or system generates a high resolution 3-D point cloud to operate an autonomous driving vehicle (ADV) from a low resolution 3-D point cloud and camera-captured image(s). The system receives a first image captured by a camera for a driving environment. The system receives a second image representing a first depth map of a first point cloud corresponding to the driving environment. The system downsamples the second image by a predetermined scale factor until a resolution of the second image reaches a predetermined threshold. The system generates a second depth map by applying a convolutional neural network (CNN) model to the first image and the downsampled second image, the second depth map having a higher resolution than the first depth map such that the second depth map represents a second point cloud perceiving the driving environment surrounding the ADV.

Method and data processing system for detecting a malicious component on an integrated circuit

A method and data processing system are provided for detecting a malicious component in a data processing system. The malicious component may be of any type, such as a hardware trojan, malware, or ransomware. In the method, a plurality of counters is used to count events in the data processing system during operation, where each event has a counter associated therewith. A machine learning model is trained a normal pattern of behavior of the data processing system using the event counts. After training, an operation of the data processing system is monitored using the machine learning model. Current occurrences of events in the data processing system are compared to the normal pattern of behavior. If a different pattern of behavior is detected, an indication, such as a flag, of the different pattern of behavior is provided.

Optimization device, guidance system, optimization method, and program

Upper-order parameters and lower-order parameters can be optimized by performing evaluation a small number of times. An optimization apparatus 10 includes: an evaluation unit 300 that performs calculation based on evaluation data, an upper-order parameter z, and a lower-order parameter x, and outputs an evaluation value indicating an evaluation on the calculation result; an optimization unit 100 that optimizes the upper-order parameter z and the lower-order parameter x; and an output unit 400 that outputs the optimized upper-order parameter z and lower-order parameter x that are obtained by repeating processing in the evaluation unit 300 and processing in the evaluation unit 300 . The optimization unit 100 learns a model for predicting evaluation values y based on combinations of the evaluation value y, the upper-order parameter z, and the lower-order parameter x, selects the upper-order parameter z to be used in evaluation performed by the evaluation unit 300 next, and determines the lower-order parameter x to be used in evaluation performed by the evaluation unit 300 next from among lower-order parameters x corresponding to the selected upper-order parameter z based on the learnt model.

Standardized form recognition method, associated computer program product, processing and learning systems

The present invention concerns a method for transforming an unstructured set of data representing a standardized form to a structured set of data. The method comprises a processing phase comprising the steps of determining a plurality of data blocks in the unstructured set of data using learning parameters determined using a plurality of samples, each data block corresponding to a visual pattern on the standardized form and being categorized to a known class, of processing data in each data block and of forming a structured set of data using the processed data from each data block, according to the class of this block.

Object Detection Model Training Method, Apparatus, and Device

In an object detection model training method, a classifier that has been trained in a first phase is duplicated to at least two copies, and in a training in a second phase, each classifier obtained through duplication is configured to detect to-be-detected objects with different sizes, and train an object detection model based on a detection result.

Apparatus and method for interactively viewing and clustering data segments from long data recordings

A method for operating a data processing system and computer readable medium causing a data processing system to execute that method are disclosed. The method includes causing the data processing system to receive a plurality of first EDSs classified into a plurality of first clusters and a first RDS for each of the plurality of first clusters and displaying, a first display for each of the first clusters and a RDS for each of the first clusters. The data processing system receives information from a user specifying one or more of the first clusters to be further clustered to arrive at a specified number of second clusters into which the specified one or more first clusters are to be classified, and performing a second clustering on the selected clusters. The method also includes displaying a second display that includes a plurality of second EDSs, classified into the second clusters.

Visualization of subimage classifications

Large digital images are classified by analyzing them at a subimage level and assigning classification probabilities to the subimages; these may be combined into a classification probability for the entire image. Classification probabilities may be visualized across the image using probabilities computed for the subimages. This enables ready identification of the image subregions upon which classification is based as well as the classification scores or probabilities associated therewith. For example, a large source image—too large to be analyzed directly by a neural network—may be decomposed into smaller subimages such as square tiles, which are sifted based on a visual criterion. The visual criterion may be image entropy, density, background percentage, or other discriminator. A neural network produces tile-level classifications that are aggregated to classify the source image, and overlapping tiles are used to create a probability map showing subimage probabilities.

Intelligent tracking system and methods and systems therefor

An intelligent tracking system generally includes one or more tracking devices, some of which may be passive tracking devices. Each passive tracking device includes one or more transceivers and is energized by an energizing signal. Some of these passive tracking devices may operate in a first communication mode or a second communication mode based on the energizing signal. Some tracking devices may include encryption modules or authentication modules. Some of these devices may incorporate a bulk acoustic wave oscillator.

Intelligent tracking system and methods and systems therefor

An intelligent tracking system generally includes one or more tracking devices, some of which may be passive tracking devices. Each passive tracking device includes one or more transceivers and is energized by an energizing signal. Some of these passive tracking devices may operate in a first communication mode or a second communication mode based on the energizing signal. Some tracking devices may include encryption modules or authentication modules. Some of these devices may incorporate a bulk acoustic wave oscillator.

Intelligent tracking system and methods and systems therefor

An intelligent tracking system generally includes one or more tracking devices, some of which may be passive tracking devices. Each passive tracking device includes one or more transceivers and is energized by an energizing signal. Some of these passive tracking devices may operate in a first communication mode or a second communication mode based on the energizing signal. Some tracking devices may include encryption modules or authentication modules. Some of these devices may incorporate a bulk acoustic wave oscillator.

Intelligent multi-scale medical image landmark detection

Intelligent multi-scale image parsing determines the optimal size of each observation by an artificial agent at a given point in time while searching for the anatomical landmark. The artificial agent begins searching image data with a coarse field-of-view and iteratively decreases the field-of-view to locate the anatomical landmark. After searching at a coarse field-of view, the artificial agent increases resolution to a finer field-of-view to analyze context and appearance factors to converge on the anatomical landmark. The artificial agent determines applicable context and appearance factors at each effective scale.

Inspection system, discrimination system, and learning data generation device

A characteristic included in data is determined with relatively high precision. An inspection system according to one aspect of the present invention acquires a plurality of learning data sets respectively including a combination of image data and correct answer data, and sets a difficulty level of determination for each of the learning data sets in accordance with a degree to which a result which is obtained by determining the acceptability of a product in the image data of each of the learning data sets by a first discriminator conforms to a correct answer indicated by the correct answer data. Besides, the inspection system constructs a second discriminator that determines the acceptability of the product by executing stepwise machine learning in which the learning data sets are utilized in ascending order of the set difficulty level.

Multiscale feature representations for object recognition and detection

Embodiments of the present invention are directed to a computer-implemented method for multiscale representation of input data. A non-limiting example of the computer-implemented method includes a processor receiving an original input. The processor downsamples the original input into a downscaled input. The processor runs a first convolutional neural network (“CNN”) on the downscaled input. The processor runs a second CNN on the original input, where the second CNN has fewer layers than the first CNN. The processor merges the output of the first CNN with the output of the second CNN and provides a result following the merging of the outputs.

Information processing system, method for controlling information processing system, program, and information recording medium

Provided is an information processing system capable of grouping objects without needing to pre-define a group associated with a sorting condition. A determination unit determines whether or not one of a first object and a second object different from the first object is included in a determination region that is set based on a position of another one of the first object and the second object. A grouping unit groups the first object and the second object based on a result of a comparison between the attribute information associated with the first object and the attribute information associated with the second object in a case where it is determined that the one of the first object and the second object is included in the determination region.

Tda enhanced nearest neighbors

A method comprises receiving a network of a plurality of nodes and a plurality of edges, each of the nodes of the plurality of nodes comprising members representative of at least one subset of initial data points, selecting a subset of the data points based on each node of the plurality of nodes, for each selected data point of the set of selected data points, determining a predetermined number of other data points that are closest in distance to that particular selected data point, grouping the selected data points into a plurality of groups based, at least in part, on the predetermined number of other data points of the set of selected data points that are closest in distance, each group of the plurality of groups including a different subset of data points, and providing a list of selected data points and the plurality of groups.

Method for image processing, non-transitory computer readable storage medium, and electronic device

A method for image processing includes the following. Edge pixel points in the image to be processed are determined by performing edge detection on the image to be processed. The number of the edge pixel points in the image to be processed is determined as a first pixel number. Determine that a moire pattern exists in the image to be processed when the first pixel number is larger than a first number threshold.

Automated bone segmentation in images

A method for automated segmentation of a bone in an image dataset of a region of a human body, wherein said region comprises at least two adjacent bones, and wherein one of the bones comprises a shaft region and a head region, the method comprising identifying in the image dataset voxels belonging to a contiguous volume comprising both bones based on a density threshold; extracting a subset comprising the head region of the first bone, based on a boundary image slice; and applying a trained machine learning classifier to said subset, to generate a classification of each of voxel as belonging to one of the bones or a none-bone tissue.

Method and apparatus for predicting flight delay, device and storage medium

The disclosure provides a method and an apparatus for predicting flight delay, a device and a storage medium. The method includes: acquiring flight historical data, where the flight historical data includes take-off amount and delay amount of flights during each of a plurality of time periods; determining prior knowledge of each of the plurality of time periods according to the take-off amount and the delay amount of the flights during each of the plurality of time periods; constructing a SVM prediction model according to the prior knowledge and a standard SVM model; and predicting a flight delay situation according to the SVM prediction model. The prediction of the flight delay situation for each of the plurality of time periods is realized.

Multiple targets-tracking method and apparatus, device and storage medium

The present disclosure provides a multiple targets-tracking method and apparatus, a device and a storage medium, wherein the method comprises: obtaining a to-be-processed current image, inputting the current image into a convolutional neural network model obtained by pre-training, and obtaining a target detection result; respectively extracting feature vectors of each detected target from a pre-selected convolutional layer; respectively calculating a similarity between the feature vectors of each target in the current image and feature vectors of each target of previous images, completing association of the same target between different image frames according to calculation results, and allocating a tracking serial number. The solution of the present disclosure may be applied to meet requirements for real-time processing.

Generalizable medical image analysis using segmentation and classification neural networks

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a final classification output for an image of eye tissue. The image is provided as input to each of one or more segmentation neural networks to obtain one or more segmentation maps of the eye tissue in the image. A respective classification input is generated from each of the segmentation maps. For each of the segmentation maps, the classification input for the segmentation map is provided as input to each of one or more classification neural networks to obtain, for each segmentation map, a respective classification output from each classification neural network. A final classification output for the image is generated from the respective classification outputs for each of the segmentation maps.

Foreground detection for video stabilization

Video foreground and background detection is performed on a received video sequence on a macroblock by macroblock basis using motion vector encoding information associated with each macroblock. Some of the macroblocks in each frame, for example, edge or corner macroblocks, may be exempt from the foreground and background detection. The detection information, along with the video data, is passed to an image stabilization algorithm that prepares the video for subsequent video compression.

Method and system for reducing localized artifacts in imaging data

A method and system for reducing localized artifacts in imaging data, such as motion artifacts and bone streak artifacts, are provided. The method includes segmenting the imaging data to identify one or more suspect regions in the imaging data near which localized artifacts are expected to occur, defining an artifact-containing region of interest in the imaging data around each suspect region, and applying a local bias field within the artifact-containing regions to correct for the localized artifacts.

Characterization of a ball game racket string pattern

The present invention relates to a method for characterizing a string pattern of a ball game racket frame as well as to the representation of a string pattern image of a strung ball game racket frame.

Systematic disposal, classification and dynamic procurement of recyclable resin

This disclosure relates method and system for systematic disposal and dynamic procurement of recyclable waste resin. Typical systems for disposal and procurement of recyclable resin pose challenges such as lack of segregation and collection of different types of plastics, data on recycling centers in proximity, uncertainty over resin identification codes, and so on. The disclosed system provides a framework having multiple smart collection units that may be communicatively coupled with a server. The smart bins are capable of collecting plastic items and taking multiple images thereof. The system determines information on type, weight, and location of the plastic resin in the plastic item by means of a trained CNN model, and stores in a repository. The system further includes a route optimization model that is enables selection of a set of collection units for procurement of the resin (of a specific type and a specific quantity).

System of joint brain tumor and cortex reconstruction

System for performing fully automatic brain tumor and tumor-aware cortex reconstructions upon receiving multi-modal MRI data (T1, T1c, T2, T2-Flair). The system outputs imaging which delineates distinctions between tumors (including tumor edema, and tumor active core), from white matter and gray matter surfaces. In cases where existing MRI model data is insufficient then the model is trained on-the-fly for tumor segmentation and classification. A tumor-aware cortex segmentation that is adaptive to the presence of the tumor is performed using labels, from which the system reconstructs and visualizes both tumor and cortical surfaces for diagnostic and surgical guidance. The technology has been validated using a publicly-available challenge dataset.

Photographing Method, Photo Management Method and Device

A photographing method that includes: acquiring to-be-photographed first content; after determining a first subject with which a user is concerned in the first content, acquiring an image composition relationship between a second subject in the first content and the first subject, where the second subject is another background subject in the first content except the first subject; matching the image composition relationship between the second subject and the first subject with a preset image composition template to obtain a matching evaluation degree, and providing an image composition adjustment suggestion on the first content for the user according to the matching evaluation degree and the image composition template, where the adjustment suggestion is a tip on how to adjust the image composition relationship in the first content so that the image composition relationship completely matches the preset image composition template.

Systems, methods, and media for artificial intelligence process control in additive manufacturing

Systems, methods, and media for additive manufacturing are provided. In some embodiments, an additive manufacturing system comprises: a hardware processor that is configured to: receive a captured image; apply a trained failure classifier to a low-resolution version of the captured image; determine that a non-recoverable failure is not present in the printed layer of the object; generate a cropped version of the low-resolution version of the captured image; apply a trained binary error classifier to the cropped version of the low-resolution version of the captured image; determine that an error is present in the printed layer of the object; apply a trained extrusion classifier to the captured image, wherein the trained extrusion classifier generates an extrusion quality score; and adjust a value of a parameter of the print head based on the extrusion quality score to print a subsequent layer of the printed object.

Content classification method and classification model generation method

A classification model which classifies contents is provided. Learning contents and contents are included. The learning contents are provided with a first feature and a learning label, and the contents are provided with a second feature. The content classification method includes a step of generating a plurality of first classification models by machine learning using the plurality of learning contents, a step of generating a second classification model with the use of the plurality of first classification models, and a step of providing judgment data for the plurality of contents with the use of the second classification model and performing display on a GUI. The judgment data includes a classification label or a score. The GUI designates a particular numerical range of the score and displays a corresponding content in a list form. Note that features provided for the contents are management parameters (metadata).

Smart waste receptacle

Embodiments relate generally to waste receptacles and more particularly to smart waste receptacles capable of identifying waste. A waste receptacle may identify waste using machine vision and perform a task in response to identifying the waste. In an embodiment, the smart waste receptacle may instruct a user to direct the waste to one of a plurality of waste bins based on a category of the waste. In an embodiment, the smart waste receptacle may place an order to replenish the stock of a product based on the identification of waste. In an embodiment, the smart waste receptacle is created by attaching a smart device to a standard waste receptacle.

Image based orthodontic treatment refinement

A method for dental treatment may include receiving photos capturing a state of a patient's dentition of a patient at a specific time, retrieving a first treatment plan to treat the patient's dentition, identifying an intended arrangement of the first treatment plan at the specific time, evaluating the photo parameters of the photos to generate alignment data to align the intended arrangement of the first treatment plan and the photos, generating an alignment mesh using the alignment data, the alignment mesh comprising a three-dimensional (3D) mesh representation of the patient's dentition at the specific time, performing an evaluation of the first treatment plan for modifications using the alignment mesh, and identifying proposed modifications to the first treatment plan based on the evaluation.

Image product creation based on face images grouped using image product statistics

A computer-implemented method for creating an image product by accurately grouping faces includes receiving an initial set of face groups for a plurality of face images, training classifiers between pairs of face groups in the initial set of face groups using image-product statistics by a computer processor, classifying the plurality of face images by classifiers to output binary vectors for the plurality of face images by the computer processor, calculating a value for an improved similarity function using the binary vectors for each pair of the plurality of face images, grouping the plurality of face images into modified face groups based on values of the binary similarity functions by the computer processor, and creating an image product based at least in part on the modified face groups

Method and system for feature extraction and decision making from series of images

Apparatus and methods comprise examination of a subject using images of the subject. The images can provide a non-invasive analysis technique and can include a plurality of images of a portion of the subject at different times a temperature stimulus applied to the subject. An image of the portion of the subject can be aligned such that each pixel of the image corresponds to the same point on the subject over a sequence of images of the portion. The sequence of images can be processed after aligning the images such that data is extracted from the images. The extracted data can be used to make decisions regarding the health status of the subject. Additional apparatus, systems, and methods are disclosed.

Method for focusing a camera

Aspects of the present disclosure are directed to a method for focusing a camera. In one embodiment, the method includes: dividing the field of view of the camera in to at least two segments: assigning, in each case, at least one operating element or at least one position of an operating element to the at least two segments; recognizing and tracking at least one object in at least two segments; automatically assigning the recognized at least two objects to the respective operating element or position of the operating element depending on which segment the objects are assigned to; and focusing the camera on the object assigned to the operating element or the position of the operating element in response to the operating element being actuated or the operating element being brought into the corresponding position.

Surface defect detecting method and surface defect detecting apparatus

A detecting method of optically detecting a surface defect of a moving steel material includes an irradiation step of irradiating an examination target part with illumination light beams from different directions by two or more distinguishable light sources whose light emission durations are set based on at least an allowable positional displacement of an image, the two or more distinguishable light sources repeatedly emitting light such that their light emission timings thereof do not overlap each other; and a detection step of obtaining images by reflected light beams of the respective illumination light beams and detecting a surface defect in the examination target part by executing subtraction processing between the obtained images.

Malfunction detection apparatus capable of detecting actual malfunctioning device not due to abnormal input values

A first classification circuit obtains first measured values from each of devices, the first measured values of the device including at least one input value to the device and at least one output value from the device, and classifies the first measured values of the devices into normal first measured values and outlier first measured values using the OCSVM (One Class nu-Support Vector Machine). A second classification circuit obtains second measured values from each of devices, the second measured values of the device including at least one input value to the device, and classifies the second measured values of the devices into normal second measured values and outlier second measured values using the OCSVM. A determination circuit determines a device having the outlier first measured values and the normal second measured values, to be a malfunctioning device.

Systems and methods for automated detection in magnetic resonance images

Some aspects include a method of detecting change in degree of midline shift in a brain of a patient. The method comprises, while the patient remains positioned within the low-field magnetic resonance imaging device, acquiring first magnetic resonance (MR) image data and second MR image data of the patient's brain; providing the first and second MR data as input to a trained statistical classifier to obtain corresponding first and second output, identifying, from the first output, at least one initial location of at least one landmark associated with at least one midline structure of the patient's brain; identifying, from the second output, at least one updated location of the at least one landmark; and determining a degree of change in the midline shift using the at least one initial location of the at least one landmark and the at least one updated location of the at least one landmark.

Generating and utilizing normalized scores for classifying digital objects

The present disclosure is directed toward systems and methods that enable more accurate digital object classification. In particular, disclosed systems and methods address inaccuracies in digital object classification introduced by variations in classification scores. Specifically, in one or more embodiments, disclosed systems and methods generate probability functions utilizing digital test objects and transform classifications scores into normalized classification scores utilizing probability functions. Disclosed systems and methods utilize normalized classification scores to more accurately classify and identify digital objects in a variety of applications.

Scan region determining apparatus

The invention relates to a scan region determining apparatus ( 12 ) for determining a scan region of a subject to be scanned by a scanning system ( 10 ) like a computed tomography system. A spatial transformation defining a registration of an overview image and a template image with respect to each other is determined, wherein initially the overview image and the template image are registered by using an element position indicator being indicative of a position of an element of the subject with respect to the overview image. A template scan region is defined with respect to the template image, wherein a final scan region is determined by projecting the template scan region onto the overview image by using the determined spatial transformation. The registration and thus the determination of the spatial transformation are very robust, which improves the quality of determining the final scan region.

Back-end application code stub generation from a front-end application wireframe

As part of identifying a theme corresponding to a wireframe, the wireframe comprising a set of graphical elements is analyzed, the set of graphical elements specifying a graphical representation of a user interface of a front-end application. A similarity measure is computed, the similarity measure quantifying a degree of similarity between the theme and an entry in a feature implementation history stored in a code repository. From the entry, a first feature to be implemented in a back-end application is extracted, the first feature servicing a data request from the front-end application. A source code stub extracted from the code repository, comprising a partial implementation of the first feature in the back-end application, is coupled with a first graphical element in the set of graphical elements in the wireframe.

Back-end application code stub generation from a front-end application wireframe

As part of identifying a theme corresponding to a wireframe, the wireframe comprising a set of graphical elements is analyzed, the set of graphical elements specifying a graphical representation of a user interface of a front-end application. A similarity measure is computed, the similarity measure quantifying a degree of similarity between the theme and an entry in a feature implementation history stored in a code repository. From the entry, a first feature to be implemented in a back-end application is extracted, the first feature servicing a data request from the front-end application. A source code stub extracted from the code repository, comprising a partial implementation of the first feature in the back-end application, is coupled with a first graphical element in the set of graphical elements in the wireframe.

Method and System for Cluster-Based Video Monitoring and Event Categorization

A computing system obtains a respective motion vector for each of a series of motion event candidates in real-time as said each motion event candidate is detected in a live video stream. In response to receiving the respective motion vector for each of the series of motion event candidates, the computing system determines a spatial relationship between the respective motion vector of said each motion event candidate to one or more existing clusters established based on a plurality of previously processed motion vectors, and in accordance with a determination that the respective motion vector of a first motion event candidate of the series of motion event candidates falls within a respective range of at least a first existing cluster of the one or more existing clusters, assigns the first motion event candidate to at least a first event category associated with the first existing cluster.

Method for Microscopic Image Acquisition Based on Sequential Section

A method for microscopic image acquisition based on a sequential slice. The method includes; acquiring a sample of the sequential slice and a navigation image thereof; identifying and labeling the sample of the sequential slice in the navigation image by utilizing methods of image processing and machine learning; placing the sample of the sequential slice in a microscope, establishing a coordinate transformation matrix for a navigation image-microscope actual sampling space coordinate, and navigating and locating a random pixel point in the navigation image to a center of the microscope's visual field; locating the sample of the sequential slice under a low resolution visual field, binding a sample acquisition parameter; based on the binding of the sample acquisition parameter, recording a relationship of relative of locations between a center point of a high resolution acquisition region and a center point after being matched with a sample template.

Methods and apparatus for acquisition and tracking, object classification and terrain inference

A target object tracking system ( 1 ) includes a processor ( 5 ) for receiving image data (S 1 ) captured by one or more sensor ( 7 ) disposed on the vehicle ( 2 ). The processor ( 5 ) is configured to analyse the image data to identify image components (IMC(n)) and to determine a movement vector (V(n)) of each image component (IMC(n)). The movement vectors each include a magnitude and a direction. At least one of the image components (IMC(n)) is classified as a target image component relating to the target object and at least one of the remaining image components (IMC(n)) as a non-target image component. The movement vector (V(n)) of the at least one target image component is modified in dependence on the movement vector of the or each non-target image component. The target object ( 3 ) is tracked in dependence on the modified movement vector of the at least one target image component. The disclosure also relates to a method and a non-transitory computer-readable medium.

System and Method for Free Space Estimation

A system and method for estimating free space including applying a machine learning model to camera images of a navigation area, where the navigation area is broken into cells, synchronizing point cloud data from the navigation area with the processed camera images, and associating probabilities that the cell is occupied and object classifications of objects that could occupy the cells with cells in the navigation area based on sensor data, sensor noise, and the machine learning model.