Patents Issued in November 12, 2020
-
Publication number: 20200356776Abstract: Systems and methods of augmenting objects associated with personal data. A system includes a communication module, a processor, and a memory. The memory stores instructions that, when executed, configure the processor to authenticate a client device based on a credential associated with an account record. The processor receives, from the client device, an indication of a document marker and obtains, from the account record, personal data associated with the document marker. The processor transmits, to the client device, display data based on the personal data. The display data configures the client device to display an augmented reality output based on the personal data and at least one further image of the document. The personal data may include dynamic data varying over time. The display data configuring the client device to display the augmented output may be based on current personal data obtained from the account record.Type: ApplicationFiled: May 10, 2019Publication date: November 12, 2020Applicant: The Toronto-Dominion BankInventors: Miguel NAVARRO, Levi SUTTER, Sadia ZAIDI, Mohamed ABBAS
-
Publication number: 20200356777Abstract: A system is provided which utilizes multiple combinations of object location technology to locate objects and direct users to them, and which provides reliable owner recognition and ownership verification with the use of displayed augmented reality with a predefined image of the object and/or the user. Further, the system utilizes augmented reality fingerprint markers. When the augmented reality fingerprint marker is positioned on an object and scanned with a smart device, information relating to the object is superimposed on the object displayed on the smart device.Type: ApplicationFiled: July 28, 2020Publication date: November 12, 2020Inventor: Carl Lamont
-
Publication number: 20200356778Abstract: A process for fixed camera and unmanned mobile device collaboration is disclosed in order to improve identification of an object of interest. A first point of view (POV) of a captured object is obtained and it is determined, with a first level of certainty, that the captured first POV of the object matches a stored object of interest. A dispatch instruction and intercept information is then wirelessly broadcast for receipt by camera-equipped unmanned mobile vehicles within a broadcast range for identifying and intercepting the object. Subsequently, a captured second POV of the first captured object is received via the one or more camera-equipped unmanned mobile vehicles. The captured second POV of the captured object is used to determine, with a second level of certainty, that the captured object matches the stored object of interest.Type: ApplicationFiled: July 30, 2020Publication date: November 12, 2020Inventors: WOJCIECH JAN KUCHARSKI, PAWEL JURZAK, GRZEGORZ KAPLITA
-
Publication number: 20200356779Abstract: Methods, systems and computer program products for flagging abnormal videos are provided. Aspects include training an image recognition model based on a plurality of images that depict one or more of a plurality of subjects. Aspects also include generating a normal subject relationship graph representing normal relationships between the plurality of subjects by applying the image recognition model to a plurality of training videos and a test subject relationship graph representing test relationships between subjects depicted in a test video by applying the image recognition model to the test video. Each normal relationship is associated with a strength value. Responsive to determining that a difference between a strength value associated with a first normal relationship and a strength value associated with a corresponding first test relationship exceeds a predetermined threshold, aspects include flagging the test video as being abnormal.Type: ApplicationFiled: May 10, 2019Publication date: November 12, 2020Inventors: Ying-Chen Yu, Chih-Wen Su, Jeff Hsueh-Chang Kuo, June-Ray Lin
-
Publication number: 20200356780Abstract: A video processing method and a terminal device are provided. The video processing method includes acquiring video data, acquiring a plurality of video segments from the video data according to flight parameter information of an unmanned aerial vehicle or motion parameter information of a capturing device, and obtaining a target video by processing the plurality of video segments according to a preset parameter.Type: ApplicationFiled: June 4, 2020Publication date: November 12, 2020Inventors: Qiheng SONG, Xinyu LI, Jianghui LIU
-
Publication number: 20200356781Abstract: The object of the present disclosure is to provide a method, apparatus and selection engine for classification matching of videos. The method according to the present disclosure includes: performing multi-dimensional identification of the content of at least one video in order to determine the abstract information of the at least one video; generating the respective classification attribute information of the at least one video based on the respective abstract information of the at least one video, wherein the classification attribute information includes the humanistic attribute information of the video; wherein the humanistic attribute information is used to indicate the value judgement corresponding to the video. The present disclosure has the following advantages: fully excavate various classification attributes of video content through multi-dimensional identification of the video, and then obtains the humanistic attributes of the video.Type: ApplicationFiled: January 9, 2019Publication date: November 12, 2020Inventors: Jiangchun Luo, Xiyan Chen
-
Publication number: 20200356782Abstract: Embodiments of this application disclose a video processing method. The method may include: obtaining bullet comment data corresponding to a video data; obtaining keyword information entry matching the bullet comment data from a key information library as target keyword information entry, the key information library comprising keyword information entries and classification recognition models of target objects respectively corresponding to each of the keyword information entries; obtaining a target video frame from a plurality of video frames of the video data; recognizing an image region of a target object corresponding to the target keyword information entry in the target video frame based on a classification recognition model of the target object; determining the recognized image region as a target region; and performing animation processing on the target region in the target video frame in response to the target video frame in the video data being played.Type: ApplicationFiled: July 23, 2020Publication date: November 12, 2020Applicant: TENCENT TECHNOLOGY (SHENZHEN) COMPANY LIMITEDInventor: Yujie LIU
-
Publication number: 20200356783Abstract: Systems, methods, and apparatus for identifying and tracking UAVs including a plurality of sensors operatively connected over a network to a configuration of software and/or hardware. Generally, the plurality of sensors monitors a particular environment and transmits the sensor data to the configuration of software and/or hardware. The data from each individual sensor can be directed towards a process configured to best determine if a UAV is present or approaching the monitored environment. The system generally allows for a detected UAV to be tracked, which may allow for the system or a user of the system to predict how the UAV will continue to behave over time. The sensor information as well as the results generated from the systems and methods may be stored in one or more databases in order to improve the continued identifying and tracking of UAVs.Type: ApplicationFiled: March 16, 2020Publication date: November 12, 2020Inventors: Rene SEEBER, Ingo Seebach, Henning Meyer, Markus Schoeler, Kai Baumgart, Christian Scheibe, David Prantl
-
Publication number: 20200356784Abstract: A state acquisition unit (2020) acquires a state of a monitoring target in a captured image captured by a camera (3040). A monitoring point acquisition unit (2040) acquires, from a monitoring point information storage unit (3020), a monitoring point corresponding to the state of the monitoring target acquired by the state acquisition unit (2020). The monitoring point indicates a position to be monitored in the captured image. A presentation unit (2060) presents the monitoring point on the captured image.Type: ApplicationFiled: May 20, 2020Publication date: November 12, 2020Inventors: Ryoma Oami, Hiroyoshi Miyano, Yusuke Takahashi, Hiroo Ikeda, Yukie Ebiyama, Ryo Kawai, Takuya Ogawa, Kazuya Koyama, Hiroshi Yamada
-
Publication number: 20200356785Abstract: A method of detecting a back of a shelf for supporting objects includes: obtaining an image depicting a shelf having a shelf edge and a support surface extending from the shelf edge to a shelf back; decomposing the image into a plurality of patches; for each patch: generating a feature descriptor; based on the feature descriptor, assigning one of a shelf back classification and a non-shelf back classification to the patch; generating a mask corresponding to the image, the mask containing an indication of the classification assigned to each of the patches; and presenting the mask.Type: ApplicationFiled: July 27, 2020Publication date: November 12, 2020Inventors: Raymond Phan, Yan Zhang, Richard Jeffrey Rzeszutek, Bo Fu
-
Publication number: 20200356786Abstract: The present disclosure relates to methods and systems to manage traffic density in a transportation system, and by doing so, maintain, in one embodiment, traffic flows near optimum levels to maximize road capacity and minimize travel times. The method includes, in one embodiment, a mechanism for vehicles to request road access from a centralized control, a queuing system that allows road access to be granted to individual vehicles over an extended period of time in a fair and organized fashion, a measurement system that allows traffic flow and density throughout the system to be determined in real-time, and an enforcement and fraud prevention mechanism to ensure that the rules and permissions imposed by the system are followed.Type: ApplicationFiled: May 22, 2020Publication date: November 12, 2020Inventor: Marc R. Hannah
-
Publication number: 20200356787Abstract: A method for determining the digital fingerprint of vehicles in transit for automatic charge of tolls, fees and/or other possible treatments, comprises the steps of: providing at least one portal along a road with at least one pair of cameras respectively facing upstream and downstream of the portal and with at least one antenna configured to interrogate vehicle on-board devices to receive an identification code therefrom; processing images and signals from the cameras and the antenna to assign each vehicle a series of data which together define a digital fingerprint of each vehicle, in particular by tracking each vehicle shot by the cameras along a track and obtaining from the images acquired along the track data which are used by a classification algorithm for classifying the vehicles.Type: ApplicationFiled: May 6, 2020Publication date: November 12, 2020Applicant: SINELEC S.p.A.Inventor: Pietro Contegno
-
Publication number: 20200356788Abstract: A vehicular trailer assist system includes a control that includes an image processor for processing image data captured by the camera representative of at least a portion of a trailer hitched to the vehicle. The control determines whether the trailer has been previously hitched to the vehicle. Responsive to the trailer not being previously hitched, the control operates in a trailer initial calibration mode. Responsive to the control recognizing the trailer, the control operates in a recognized trailer calibration mode. The control obtains calibration data unique to the hitched trailer. The control, responsive to obtaining the calibration data, processes image data captured by the camera using the calibration data to locate the current position of the trailer relative to the vehicle. The control, responsive to locating the current position of the trailer relative to the vehicle, determines a trailer angle based on the located current position.Type: ApplicationFiled: May 8, 2020Publication date: November 12, 2020Inventors: Harold E. Joseph, Joshua Teichroeb, Alexander Velichko, Jyothi P. Gali, Guruprasad Mani Iyer Shankaranarayanan
-
Publication number: 20200356789Abstract: A shading device comprises: a shading member; a display apparatus disposed on a surface of the shading member in such a manner that a display portion faces to an operator; an image pickup device to pick up, as an image, a region which an opposite surface of the surface faces, and generate image pickup data; and a data processing circuit to generate display image data, based on the image pickup data. An image display module comprises: a display apparatus to be disposed on a surface of a shading device in such a manner that a display portion faces an operator; an image pickup device to pick up, as an image, a region to which an opposite surface of the surface faces, and generate image pickup data; and a data processing circuit to generate display image data, based on the image pickup data.Type: ApplicationFiled: July 23, 2020Publication date: November 12, 2020Inventor: Katsuhiko Kishimoto
-
Publication number: 20200356790Abstract: A computer, including a processor and a memory, the memory including instructions to be executed by the processor to generate a pair of synthetic stereo images and a corresponding synthetic depth map with an image synthesis engine wherein the synthetic stereo images correspond to real stereo images acquired by a stereo camera and the synthetic depth map is a three-dimensional (3D) map corresponding to a 3D scene viewed by the stereo camera and process each image of the pair of synthetic stereo images pair independently using a generative adversarial network (GAN) to generate a fake image, wherein the fake image corresponds to one of the synthetic stereo images.Type: ApplicationFiled: May 8, 2019Publication date: November 12, 2020Applicant: Ford Global Technologies, LLCInventors: Nikita Jaipuria, Gautham Sholingar, Vidya Nariyambut Murali
-
Publication number: 20200356791Abstract: A method for image processing at different illumination conditions, the method may include acquiring an image of an environment of a vehicle; selecting a set of pixels located within a region of interest that is located at an upper part of the image; calculating an illumination condition indicator based on values of the set of pixels; selecting a selected machine learning process, out of a machine learning processes, based on the illumination condition indicator; wherein different machine learning processes are trained to different illumination conditions; and processing the image by the selected machine learning process to provide processing results.Type: ApplicationFiled: February 25, 2020Publication date: November 12, 2020Applicant: Cartica AI LtdInventors: Igal Raichelgauz, Karina Odinaev
-
Publication number: 20200356792Abstract: A solid object detection device includes an overhead view transformation processing unit transforming first and second photographed images photographed by a camera at different timings in travel of a vehicle into first and second overhead view images, respectively, a subtracted image generation unit generating a subtracted image between the first and second overhead view images whose photographing positions are aligned with each other, a solid object position specification unit specifying a position of a solid object present around the vehicle based on the subtracted image, and a masked subtracted image generation unit generating a masked subtracted image in which a region other than a solid object candidate region as a candidate where the solid object appears in the subtracted image is masked and the solid object position specification unit specifies a position of the solid object in the subtracted image based on the masked subtracted image.Type: ApplicationFiled: April 30, 2020Publication date: November 12, 2020Applicant: CLARION CO., LTD.Inventors: Ayano MIYASHITA, Naoki SHIMIZU, Hiroaki ANDOU, Kengo ASAKI
-
Publication number: 20200356793Abstract: A method for detecting parking violation associated with a vehicle is provided and includes: after an engine unit of the vehicle is switched to an activated state, controlling an image capturing unit to continuously capture images of a surrounding environment of the vehicle; determining whether the vehicle is in a stationary state; when it is determined that the vehicle is in a stationary state, performing an image processing procedure on at least one of the images for determining whether a violation condition is met, the violation condition indicating parking violation of the vehicle; and when the determination is affirmative, generating an alert for output.Type: ApplicationFiled: May 7, 2020Publication date: November 12, 2020Inventor: Jui-Ying CHUNG
-
Publication number: 20200356794Abstract: Aspects of the disclosure relate to training and using a phrase recognition model to identify phrases in images. As an example, a selected phrase list may include a plurality of phrases is received. Each phrase of the plurality of phrases includes text. An initial plurality of images may be received. A training image set may be selected from the initial plurality of images by identifying the phrase-containing images that include one or more phrases from the selected phrase list. Each given phrase-containing image of the training image set may be labeled with information identifying the one or more phrases from the selected phrase list included in the given phrase-containing images. The model may be trained based on the training image set such that the model is configured to, in response to receiving an input image, output data indicating whether a phrase of the plurality of phrases is included in the input image.Type: ApplicationFiled: May 20, 2020Publication date: November 12, 2020Inventors: Victoria Dean, Abhijit S. Ogale, Henrik Kretzschmar, David Harrison Silver, Carl Kershaw, Pankaj Chaudhari, Chen Wu, Congcong Li
-
Publication number: 20200356795Abstract: A movable carrier auxiliary system includes a driver state detecting device and a control device. The driver state detecting device includes a biometric feature detecting module, a storage module, and an operation module. The biometric feature detecting module detects a biometric feature of a driver. The storage module stores a first biometric feature parameter, a second biometric feature parameter, a first operating mode corresponding to the first biometric feature parameter, and a second operating mode corresponding to the second biometric feature parameter. The operation module detects whether the biometric feature of the driver matches with the first biometric feature parameter or the second biometric feature parameter or not via the biometric feature detecting module, and to correspondingly generate a detection signal. The control device retrieve the first operating mode or the second biometric feature parameter from the storage module to control the movable carrier based on the detection signal.Type: ApplicationFiled: March 19, 2020Publication date: November 12, 2020Applicant: ABILITY OPTO-ELECTRONICS TECHNOLOGY CO., LTD.Inventors: YEONG-MING CHANG, Chien-Hsun Lai, Yao-Wei Liu
-
Publication number: 20200356796Abstract: A user establishes an account with an account management system and downloads a service application on a user computing device associated with the user. The user enters a service provider location and signs into a service application. A service camera device captures a video feed of one or more users within a visual field of the service camera device and a service computing device compares movement data received from user computing devices at the service provider location against movement of detected objects in the video feed to identify one or more users in the video feed. After identifying the one or more users within the video feed of the service camera device, in response to a particular user initiating a service request at the service computing device, the service computing device identifies the particular user at the service computing device as being within a service area of the video feed.Type: ApplicationFiled: July 23, 2020Publication date: November 12, 2020Inventor: Phillip Ellsworth Stahlfeld
-
Publication number: 20200356797Abstract: In a verification method for verifying a verification target camera, an original image in which optical characteristics of an original image camera different from the verification target camera are corrected to a captured image of a real scene captured by the original image camera is prepared, a verification image in which an influence of optical characteristics of the verification target camera is applied to the original image is prepared, image recognition of the verification image is performed by executing a verification target algorithm applied for image recognition of an image captured by the verification target camera, and an image recognition result of the verification image is evaluated.Type: ApplicationFiled: May 1, 2020Publication date: November 12, 2020Inventor: Toshiyuki TSUCHIMOTO
-
Publication number: 20200356798Abstract: In an illustrative embodiment, a system for identifying products on a production line includes image capturing devices that acquire images of containers moving along a production line at an inspection location. The system also includes a rejection device and a controller that configures the image capturing devices for image acquisition based on properties of the containers, identifies a product associated with each of the containers based on a portion of a product identification code and a portion of additional features detected in the images, and determines whether the identified product matches predetermined properties or characteristics, resulting in a pass result, otherwise a non-pass result occurs. When a non-pass result occurs, the controller outputs a signal to actuate the rejection device that removes the container from the production line.Type: ApplicationFiled: July 27, 2020Publication date: November 12, 2020Applicant: Sensors IncorporatedInventor: David J. Kotula
-
Publication number: 20200356799Abstract: The system determines whether content such as an image is suitable for content modification based on one or more criteria. The system includes decision engines or modules configured to evaluate one or more suitability metrics based on corresponding criteria such as publication status, restriction status, context, compatibility, and classification. If content is unsuitable for content modification because of entities or context depicted therein, privacy status, incompatibility with content modification, properties of the content file itself, or other aspects, the system generates a tag indicating the content is unsuitable for content modification. If content is suitable for content modification because of entities or context depicted therein, publication status, compatibility with content modification, properties of the content file itself, or other aspects, the system generates a content modification tag indicating the content is suitable for content modification.Type: ApplicationFiled: May 6, 2019Publication date: November 12, 2020Inventor: Alejandro Sanchez Pulido
-
Publication number: 20200356800Abstract: Various embodiments provide a polygonal region detection method and apparatus, a computer readable storage medium and an electronic device. In those embodiments, a to-be-detected image can be obtained. A plurality of line segments in the image can be calculated based on a line detection algorithm. The plurality of line segments meeting a merging condition can be merged into a line segment. Crosspoints of the pairwise merged line segments can be calculated according to the merged line segments in the image. A polygonal region can be generated with the crosspoints as vertexes of the polygonal region in the image.Type: ApplicationFiled: October 24, 2018Publication date: November 12, 2020Inventors: Beier ZHU, Rui ZHANG
-
Publication number: 20200356801Abstract: The present disclosure provides an object detection method and an object detection device. The object detection device includes: a heterogeneous processor and a memory, the heterogeneous processor including: a processing unit and a programmable logic unit, wherein the programmable logic unit is configured to receive a to-be-detected image, perform feature extraction on the to-be-detected image, and write an extracted feature into the memory; the processing unit is configured to read the feature from the memory, perform target object detection according to the feature, and output a detection result to the programmable logic unit; and the programmable logic unit is further configured to receive the detection result, generate prompt information according to the detection result, and output the prompt information.Type: ApplicationFiled: September 17, 2019Publication date: November 12, 2020Inventor: Ran DUAN
-
Publication number: 20200356802Abstract: Embodiments of the present application provide an image processing method and apparatus, an electronic device, a storage medium, and a program product. The method includes: generating a feature map of a to-be-processed image by performing feature extraction on the image; determining a feature weight corresponding to each of a plurality of feature points comprised in the feature map; and obtaining a feature-enhanced feature map by separately transmitting feature information of each feature point to associated other feature points comprised in the feature map based on the corresponding feature weight.Type: ApplicationFiled: June 18, 2020Publication date: November 12, 2020Inventors: Hengshuang ZHAO, Yi ZHANG, Jianping SHI
-
Publication number: 20200356803Abstract: Systems and methods for integrated multi-factor multi-label analysis include using one or more deep learning systems, such as neural networks, to analyze how well one or more entities are likely to benefit from a targeted action. Data associated with each of the entities is analyzed to determine a score for each of the proposed targeted actions using multiple analysis factors. The scores for each analysis factor are determined using a different multi-layer analysis network for each analysis factor. The scores for each analysis factor are then combined to determine an overall score for each of the proposed targeted actions. The entities and the proposed targeted actions with the highest scores are then identified and then used to determine which entities are to be the subject of which targeted actions.Type: ApplicationFiled: May 10, 2019Publication date: November 12, 2020Inventors: Yaqin Yang, Nitin Sharma, Fransisco Kurniadi, Yang Wu
-
Publication number: 20200356804Abstract: Provided is an image recognition device. The image recognition device includes a frame data change detector that sequentially receives a plurality of frame data and detects a difference between two consecutive frame data, an ensemble section controller that sets an ensemble section in the plurality of frame data, based on the detected difference, an image recognizer that sequentially identifies classes respectively corresponding to a plurality of section frame data by applying different neural network classifiers to the plurality of section frame data in the ensemble section, and a recognition result classifier that sequentially identifies ensemble classes respectively corresponding to the plurality of section frame data by combining the classes in the ensemble section.Type: ApplicationFiled: May 8, 2020Publication date: November 12, 2020Inventors: Ju-Yeob KIM, Byung Jo KIM, Seong Min KIM, Jin Kyu KIM, Ki Hyuk PARK, Mi Young LEE, Joo Hyun LEE, Young-deuk JEON, Min-Hyung CHO
-
Publication number: 20200356805Abstract: This application provides an image recognition method, a storage medium, and a computer device. The method includes: obtaining a to-be-recognized image; preprocessing the to-be-recognized image, to obtain a preprocessed image; obtaining, through a first submodel in a machine learning model, a first image feature corresponding to the to-be-recognized image, and obtaining, through a second submodel in the machine learning model, a second image feature corresponding to the preprocessed image; and determining, according to the first image feature and the second image feature, a first probability that the to-be-recognized image belongs to a classification category corresponding to the machine learning model. It may be seen that, the solutions provided by this application can improve recognition efficiency and accuracy.Type: ApplicationFiled: July 29, 2020Publication date: November 12, 2020Inventors: Xing SUN, Yi ZHANG, Xinyang JIANG, Xiaowei GUO, Xuan ZHOU, Jia CHANG
-
Publication number: 20200356806Abstract: Embodiments of the present invention relate to methods, systems, and computer program products for container image management. In a method, an image layer in a container image may be received by one or more processors, and the container image is to be stored in an image server comprising a group of image layers. A base portion may be selected by one or more processors from the group of image layers based on a similarity analysis between the image layer and the group of image layers. A patch portion may be generated by one or more processors based on a difference between the image layer and the selected base portion. With these embodiments, the container image may be stored based on multiple image layers, and thus the container image may be maintained in a much finer granularity so as to reduce requirements on the bandwidth and time cost for transmitting the container image.Type: ApplicationFiled: May 7, 2019Publication date: November 12, 2020Inventors: Guang Cheng LI, Qi Ming TENG, Yong ZHENG, Lin Feng SHEN
-
Publication number: 20200356807Abstract: A training device and a training method for training a multi-goal model based on goals in a goal space are provided. The training device includes a memory and a processor coupled to the memory. The processor is configured to set the goal space, to acquire a plurality of sub-goal spaces of different levels of difficulty; change a sub-goal space to be processed from a current sub-goal space to a next sub-goal space of a higher level of difficulty; select, as sampling goals, goals at least from the current sub-goal space, and to acquire transitions related to the sampling goals by executing actions; train the multi-goal model based on the transitions, and evaluate the multi-goal model by calculating a success rate for achieving goals in the current sub-goal space.Type: ApplicationFiled: May 7, 2020Publication date: November 12, 2020Applicant: FUJITSU LIMITEDInventors: Chaoliang Zhong, Wensheng Xia, Ziqiang Shi, Jun Sun
-
Publication number: 20200356808Abstract: The present disclosure relates to unsupervised visual attribute transfer through reconfigurable image translation. One aspect of the present disclosure provides a system for learning the transfer of visual attributes, including an encoder, converter and generator. The encoder encodes an original source image to generate a plurality of attribute values that specify the original source image, and to encode an original reference image to generate a plurality of attribute values that specify the original reference image. The converter replaces at least one attribute value of an attribute that is target attribute of the attribute values of the original source image with at least one corresponding attribute value of the original reference image, to obtain a plurality of attribute values that specify a target image of interest. The generator generates a target image based on the attribute values of the target image of interest.Type: ApplicationFiled: July 24, 2020Publication date: November 12, 2020Inventors: Taeksoo KIM, Byoungjip KIM, Jiwon KIM, Moonsu CHA
-
Publication number: 20200356809Abstract: Batch processing of artificial intelligence data can offer advantages, such as increased hardware utilization rates and parallelism for efficient parallel processing of data. However, batched processing in some cases can increase memory usage if batching is done without regards for its memory costs. For example, memory usage associate with batched-backpropagation can be substantial, thereby reducing desirable locality of processing data. System resources can be spent loading and traversing data inefficiently over the chip area. Disclosed are systems and methods for intelligent batching which utilizes a flexible pipelined forward and/or backward propagation to take advantage of parallelism in data, while maintaining desirable locality of data by reducing memory usage during forward and backward passes through a neural network or other AI processing tasks.Type: ApplicationFiled: May 7, 2019Publication date: November 12, 2020Inventor: Tapabrata Ghosh
-
Publication number: 20200356810Abstract: A method of training a generator G of a Generative Adversarial Network (GAN) includes generating a real contextual data set {x1, . . . , xN} for a high resolution image Y; generating a generated contextual data set {g1, . . . , gN} for a generated high resolution image G(Z); calculating a perceptual loss Lpcept value using the real contextual data set {x1, . . . , xN} and the generated contextual data set {g1, . . . , gN}; and training the generator G using the perceptual loss Lpcept value. The generated high resolution image G(Z) is generated by the generator G of the GAN in response to receiving an input Z, where the input Z is a random sample that corresponds to the high resolution image Y.Type: ApplicationFiled: August 5, 2019Publication date: November 12, 2020Inventors: Sheng Zhong, Shifu Zhou
-
Publication number: 20200356811Abstract: Computer vision systems and methods for machine learning using a set packing framework are provided. A minimum weight set packing (“MWSP”) framework is parameterized by a set of possible hypotheses, each of which is associated with a real valued cost that describes the sensibility of the belief that the members of the hypothesis correspond to a common cause. Using MWSP, the system then selects the lowest total cost set of hypotheses, such that no two selected hypotheses share a common observation. Observations that are not included in any selected hypothesis, define the set of false observations can be thought of as false observations/noise. The system can be utilized to support one or more trained computer models in performing computer vision on input data in order to generate output data.Type: ApplicationFiled: May 8, 2020Publication date: November 12, 2020Applicant: Insurance Services Office, Inc.Inventors: Julian Yarkony, Yossiri Adulyasak, Maneesh Kumar Singh, Guy Desaulniers
-
Publication number: 20200356812Abstract: The present disclosure is directed to a method and a system which trains object detection neural networks with a dependency based loss function for capturing dependent training images. The object detection neural networks system comprises a calibrated camera system for capturing images for the object detection neural network model and the dependent based loss function to process dependent training images, which is then fed to an optimizer to adjust parameters of the object detection neural network model to minimize the loss value. Additional penalties can be imposed by knowledge base rules. A camera system in the object detection neural networks system can include cameras with a fixed distance between neighboring cameras, or unaligned cameras arranged at various distances and/or angles between them, with an option to add sensors to the camera system.Type: ApplicationFiled: May 9, 2020Publication date: November 12, 2020Inventors: Mark Oleynik, Alexey Chaykin
-
Publication number: 20200356813Abstract: Aspects of the detailed technologies concern training and use of neural networks for fine-grained classification of large numbers of items, e.g., as may be encountered in a supermarket. Mitigating false positive errors is an exemplary area of emphasis. Novel network topologies are also detailed—some employing recognition technologies in addition to neural networks. A great number of other features and arrangements are also detailed.Type: ApplicationFiled: May 21, 2020Publication date: November 12, 2020Inventors: Ravi K. Sharma, Tomas Filler, Utkarsh Deshmukh, Vahid Sedighianaraki, William Y. Conwell
-
Publication number: 20200356814Abstract: Provided is an object detection device or the like which efficiently generates good-quality training data. This object detection device is provided with: a detection unit which uses a dictionary to detect objects from an input image; a reception unit which displays, on a display device, the input image accompanied by a display emphasizing partial areas of detected objects, and receives, from one operation of an input device, a selection of a partial area and an input of the class of the selected partial area; a generation unit which generates training data from the image of the selected partial area and the inputted class; and a learning unit which uses the training data to learn the dictionary.Type: ApplicationFiled: July 24, 2020Publication date: November 12, 2020Applicant: NEC CORPORATIONInventor: Yusuke TAKAHASHI
-
Publication number: 20200356815Abstract: Parallel training of a machine learning model on a computerized system is described. Computing tasks of a system can be assigned to multiple workers of the system. Training data can be accessed. The machine learning model is trained, whereby the training data accessed are dynamically partitioned across the workers of the system by shuffling subsets of the training data through the workers. As a result, different subsets of the training data are used by the workers over time as training proceeds. Related computerized systems and computer program products are also provided.Type: ApplicationFiled: May 7, 2019Publication date: November 12, 2020Inventors: Nikolas Ioannou, Celestine Duenner, Thomas Parnell
-
Publication number: 20200356816Abstract: An association metric for record attributes associated with cardinalities that are not necessarily the same is used for training and/or applying an entity resolution (ER) model. A pair of records includes (a) a first record indicating a first set of values for a first attribute and (b) a second record indicating a second set of values for a second attribute. Each of the first set of values and each of the second set of values are compared to determine individual association metrics. A first-level reduction operation is applied to subsets of the individual association metrics to determine reduced association metrics. A second-level reduction operation is applied to the reduced association metrics to determine an association metric, for the pair of records, for training and/or applying an ER model.Type: ApplicationFiled: May 8, 2019Publication date: November 12, 2020Applicant: KOMODO HEALTHInventors: Benjamin James Campbell Blalock, Alexander Graham Glenday, Jason Richard Prestinario
-
Publication number: 20200356817Abstract: Various embodiments are generally directed to techniques for embedding a data object into a multidimensional frame, such as for training an autoencoder to generate latent space representations of the data object based on the multidimensional frame, for instance. Additionally, in one or more embodiments latent space representations of data objects may be classified, such as with a machine learning algorithm. Some embodiments are particularly directed to embedding a data object comprising a plurality of object entries into a three-dimensional (3D) frame.Type: ApplicationFiled: November 1, 2019Publication date: November 12, 2020Applicant: Capital One Services, LLCInventors: Austin Grant WALTERS, Jeremy Edward GOODSITT, Mark Louis WATSON, Anh TRUONG
-
Publication number: 20200356818Abstract: Disclosed herein are techniques for detecting logos in images or video. In one embodiment, one or more candidate regions are detected for determining logos in an image. A logo is determined to be the logo in the candidate region based on matching a feature vector of a candidate region to a feature vector of the logo.Type: ApplicationFiled: July 24, 2020Publication date: November 12, 2020Inventors: Bruno Fidel Maciel Attorre, Nicolas Huynh Thien
-
Publication number: 20200356819Abstract: Provided is a computer-implemented method for determining at least one class, including the steps of: providing at least one input data set with a plurality of performance metrics; preprocessing the at least one input data set into at least one respective processed input data set with a plurality of processed performance metrics; and determining the at least one class using machine learning on the basis of the at least one processed input data set. Further, a corresponding computer program product and system is provided.Type: ApplicationFiled: April 30, 2020Publication date: November 12, 2020Inventors: Bernhard Kempter, Reiner Schmid
-
Publication number: 20200356820Abstract: Embodiments may include a method to estimate motion data based on test image data sets. The method may include receiving a training data set comprising a plurality of training data elements. Each element may include an image data set and a motion data set. The method may include training a machine learning model using the training data set, resulting in identifying one or more parameters of a function in the machine learning model based on correspondences between the image data sets and the motion data sets. The method may further include receiving a test image data set. The test image data set may include intensities of pixels in a deep-tissue image. The method may include using the trained machine learning model and the test image data set to generate output data for the test image data set. The output data may characterize motion represented in the test image data set.Type: ApplicationFiled: July 27, 2020Publication date: November 12, 2020Applicant: Verily Life Sciences LLCInventors: Eden Rephaeli, Daniele Piponi, Chinmay Belthangady, Seung Ah Lee
-
Publication number: 20200356821Abstract: Disclosed are a method, terminal and computer readable storage medium for image classification. The method includes: determining an image feature vector of an image based on a convolutional neural network, where the image comprises textual information; determining a text feature vector based on the textual information and an embedded network; determining an image-text feature vector by joining the image feature vector with the text feature vector; and determining a category of the image based on a result of a deep neural network, where the result is determined based on the image feature vector, the text feature vector and the image-text feature vector.Type: ApplicationFiled: July 17, 2020Publication date: November 12, 2020Applicant: Beijing Dajia Internet Information Technology Co., Ltd.Inventors: Zhiwei Zhang, Fan Yang
-
Publication number: 20200356822Abstract: A method, apparatus, and program product perform microstructure analysis of a digital image of rock using a trained convolutional neural network model to generate a plurality of rock features. The rock features can represent a pore space in the microstructure of the rock including pores and throats. In many implementations, a statistical process can be applied to the rock features to generate characteristics of the pore space which can be used in classifying the rock.Type: ApplicationFiled: May 6, 2019Publication date: November 12, 2020Inventors: Alexander Starostin, Alexander Nadeev
-
Publication number: 20200356823Abstract: Disclosed are a system, apparatus and techniques for evaluating a dataset to confirm that the data in the dataset satisfies a data quality metric. A machine learning engine or the like may evaluate text strings within the dataset may be of arbitrary length and encoded according to an encoding standard. Data vectors of a preset length may be generated from the evaluated text strings using various techniques. Each data vector may be representative of the content of the text string and a category may be assigned to the respective data vector. The category assigned to each data vectors may be evaluated with respect to other data vectors in the dataset to determine compliance with a quality metric. In the case that a number of data vectors fail to meet a predetermined quality metric, an alert may be generated to mitigate any system errors that may result from unsatisfactory data quality.Type: ApplicationFiled: May 8, 2019Publication date: November 12, 2020Applicant: Capital One Services, LLCInventor: Robin Astrid Epp NEUFELD
-
Publication number: 20200356824Abstract: Disclosed are a system, apparatus and techniques for evaluating a dataset to confirm that the data in the dataset satisfies a data quality metric. A machine learning engine or the like may evaluate text strings within the dataset may be of arbitrary length and encoded according to an encoding standard. Data vectors of a preset length may be generated from the evaluated text strings using various techniques. Each data vector may be representative of the content of the text string and a category may be assigned to the respective data vector. The category assigned to each data vectors may be evaluated with respect to other data vectors in the dataset to determine compliance with a quality metric. In the case that a number of data vectors fail to meet a predetermined quality metric, an alert may be generated to mitigate any system errors that may result from unsatisfactory data quality.Type: ApplicationFiled: October 15, 2019Publication date: November 12, 2020Applicant: Capital One Services, LLCInventor: Robin Astrid Epp NEUFELD
-
Publication number: 20200356825Abstract: An electronic medical record (EMR) analysis machine automatically clusters electronic medical records to produce an initial EMR analysis model and to identify high-value EMR documents such that human analysts can focus effort on labeling only high-value EMR documents to iteratively and extremely efficiently train an EMR analysis model. High-value sample EMR documents are identified as those whose membership in one or more clusters is most ambiguous, i.e., nearest the cluster boundary.Type: ApplicationFiled: May 7, 2020Publication date: November 12, 2020Inventors: John Zhu, Noah Lieberman, Ha Pham, Vishnuvyas Sethumadhavan