Patents Examined by Sean T Motsinger
-
Patent number: 11429842Abstract: A computing system is provided. The computing system includes a processor configured to execute a convolutional neural network that has been trained, the convolutional neural network including a backbone network that is a concatenated pyramid network, a plurality of first head neural networks, and a plurality of second head neural networks. At the backbone network, the processor is configured to receive an input image as input and output feature maps extracted from the input image. The processor is configured to: process the feature maps using each of the first head neural networks to output corresponding keypoint heatmaps; process the feature maps using each of the second head neural networks to output corresponding part affinity field heatmaps; link the keypoints into one or more instances of virtual skeletons using the part affinity fields; and output the instances of the virtual skeletons.Type: GrantFiled: April 26, 2019Date of Patent: August 30, 2022Assignee: Microsoft Technology Licensing, LLCInventors: Lijuan Wang, Kevin Lin, Zicheng Liu, Kun Luo
-
Patent number: 11423702Abstract: An object recognition apparatus and method are provided. The apparatus includes a processor configured to verify a target image using an object model and based on reference intermediate data extracted by a partial layer of the object model as used in an object recognition of an input image, in response to a failure of a verification of the input image after a success of the object recognition of the input image, and perform an additional verification of the target image in response to the target image being verified in the verifying of the target image.Type: GrantFiled: November 5, 2020Date of Patent: August 23, 2022Assignee: Samsung Electronics Co., Ltd.Inventors: SeungJu Han, SungUn Park, JaeJoon Han, Jinwoo Son, ChangYong Son, Minsu Ko, Jihye Kim, Chang Kyu Choi
-
Patent number: 11417030Abstract: A method for coding information of a point cloud comprises obtaining the point cloud including a set of points in a three-dimensional space; partitioning the point cloud into a plurality of objects and generating occupancy information for each of the plurality of objects; and encoding the occupancy information by taking into account the distance between the plurality of objects.Type: GrantFiled: October 26, 2020Date of Patent: August 16, 2022Assignee: TENCENT AMERICA LLCInventors: Xiang Zhang, Wen Gao, Shan Liu
-
Patent number: 11416772Abstract: Embodiments of the present disclosure include a computer-implemented method, a system, and a computer program product for integrating bottom-up segmentation techniques into a semi-supervised image segmentation machine learning model. The computer implemented method includes training a machine learning model with a labeled dataset. The labeled dataset includes ground truth segmentation labels for each sample in the labeled dataset. The computer implemented method also includes generating a pseudo labeled dataset by applying an unlabeled dataset to the machine learning model using a top-down segmentation grouping rule. The computer implemented method further includes evaluating the pseudo labeled dataset using a bottom-up segmentation grouping rule to produce evaluation results, combining the pseudo labeled dataset with the second pseudo labeled dataset into a training dataset, and then retraining the machine learning model with the training dataset.Type: GrantFiled: December 2, 2019Date of Patent: August 16, 2022Assignee: International Business Machines CorporationInventors: Hongzhi Wang, Alexandros Karargyris, Tanveer Fathima Syeda-Mahmood, Joy Tzung-yu Wu
-
Patent number: 11403488Abstract: A method for extracting information from a table includes steps as follows. Characters of a table are extracted. The characters are merged into n-gram characters. The n-gram characters are merged into words and text lines through a two-stage GNN mode. The two-stage GNN mode comprises sub steps as: spatial features, semantic features, CNN image features are extracted from a target source; a first GNN stage is processed to output graph embedding spatial features from the spatial features; and a second GNN stage is processed to output graph embedding semantic features and graph embedding CNN image features from the semantic features and the CNN image features, respectively. The text lines are merged into cells. The cells are grouped into rows, columns, and key-value pairs based on one or more adjacency matrices, a row relationship among the cells, a column relationship among the cells, and a key-value relationship among the cells.Type: GrantFiled: March 19, 2020Date of Patent: August 2, 2022Assignee: Hong Kong Applied Science and Technology Research Institute Company LimitedInventors: Wai Kai Arvin Tang, Jinning Yang
-
Patent number: 11399779Abstract: Embodiments discussed herein facilitate system-independent quantitative perfusion measurements. One example embodiment is a method, comprising: accessing 4D (Four Dimensional) perfusion imaging data of a tissue, where the 4D perfusion imaging data comprises a plurality of 3D (Three Dimensional) stacks of perfusion imaging data over time; performing at least one of artifact reduction or post-processing on the 4D perfusion image data to generate processed 4D perfusion image data; computing one or more quantitative perfusion parameters for the tissue based at least in part on the processed 4D perfusion image data; and outputting a visual representation of the one or more quantitative perfusion parameters.Type: GrantFiled: March 18, 2019Date of Patent: August 2, 2022Assignee: Case Western Reserve UniversityInventors: David L. Wilson, Brendan L. Eck, Jacob Levi, Hao Wu, Rachid Fahmi
-
Patent number: 11388355Abstract: Devices, methods, and non-transitory program storage devices are disclosed herein to provide improved multi-spectral image processing techniques for generating an enhanced output image, the techniques comprising: obtaining an N-channel (e.g., multispectral) input image; determining fusion weights and fallback weights (e.g., relative intensity weights) for each of the N-channels of the input image; blending the fusion and fallback weights based on an amount of gradient information to generate blended weights; modulating the blended weights for a plurality of frequency band representations of the input image; applying the modulated blended weights to the corresponding frequency band representations of the input image to generate a plurality of output image frequency band representations; producing an output luma image, based on the plurality of output image frequency band representations; and generating an output RGB image, based on the output luma image, which may then, e.g.Type: GrantFiled: June 12, 2020Date of Patent: July 12, 2022Assignee: Apple Inc.Inventors: Alex Hayes, Ilya Romanenko
-
Patent number: 11386535Abstract: The disclosure provides an image blending method. The method includes projecting a first image and a second image onto a projection surface by a first projector and a second projector, respectively, and the first image and the second image overlap each other; and projecting the first control pattern onto the first image, adjusting the first control pattern such that a first control pattern frame matches a boundary of an overlapping area, and thereby identifying a position of a first non-overlapping area in the first image. Additionally, a similar operation is performed by the second projector to identify a second non-overlapping area in the second image. All pixels in the first non-overlapping area and the second non-overlapping area are adjusted such that the black-level brightness of the first non-overlapping area and the second non-overlapping area correspond to the black-level brightness of the overlapping area to obtain better brightness uniformity.Type: GrantFiled: October 25, 2019Date of Patent: July 12, 2022Assignee: Coretronic CorporationInventors: Jyun-Lin Cian, Chi-Wei Lin, Chien-Chun Peng, Yung-Chiao Liu
-
Patent number: 11386302Abstract: Systems, methods, and computer program products for performing semi-supervised contrastive learning of visual representations are provided. For example, the present disclosure provides systems and methods that leverage particular data augmentation schemes and a learnable nonlinear transformation between the representation and the contrastive loss to provide improved visual representations. Further, the present disclosure also provides improvements for semi-supervised contrastive learning.Type: GrantFiled: September 11, 2020Date of Patent: July 12, 2022Assignee: GOOGLE LLCInventors: Ting Chen, Simon Kornblith, Mohammad Norouzi, Geoffrey Everest Hinton, Kevin Jordan Swersky
-
Patent number: 11380127Abstract: The present invention relates to a method of segmenting an input image representing at least one biometric trait, by means of a convolutional neural network, CNN, the method being characterized in that it comprises the implementation, by data processing means (21) of a client (2), of steps of (b) generating, by means of a first subnetwork of said CNN, referred to as a specialization block, a confidence mask associated with a ridge orientation map of the at least one biometric trait represented by said input image, referred to as a RFM mask; said RFM mask being applied to at least the input image so as to obtain a focused input image; (c) segmenting, by means of a second subnetwork of said CNN, referred to as a segmentation block, the focused input image.Type: GrantFiled: October 22, 2020Date of Patent: July 5, 2022Assignee: IDEMIA IDENTITY & SECURITY FRANCEInventors: Anthony Cazasnoves, Emilie Niaf, Laurent Kazdaghli
-
Patent number: 11367296Abstract: A layout analysis method, an electronic device, and a non-transitory computer-readable storage medium are provided. The layout analysis method includes: obtaining coordinate information of a plurality of text lines in an image; creating a layout model of the image according to the coordinate information; analyzing a layout structure of the text lines based on the layout model; and determining an order of the text lines relative to each other based on the layout structure.Type: GrantFiled: June 17, 2021Date of Patent: June 21, 2022Assignee: NEXTVPU (SHANGHAI) CO., LTD.Inventors: Linfeng Li, Haijiao Cai, Xinpeng Feng, Ji Zhou
-
Patent number: 11367227Abstract: A method of imaging includes obtaining projection data for an object representing an intensity of radiation detected along a plurality of rays through the object, obtaining an outline of the object via a secondary imaging system, the secondary imaging system using non-ionizing radiation, determining, based on the outline, a model and model parameters for the object, calculating, based on the model and the model parameters, a volumetric attenuation map for the object, and reconstructing, based on the projection data and the volumetric attenuation map, an attenuation-corrected volumetric image.Type: GrantFiled: July 28, 2020Date of Patent: June 21, 2022Assignee: CANON MEDICAL SYSTEMS CORPORATIONInventors: Karthikayan Balakrishnan, Kent C. Burr
-
Patent number: 11354778Abstract: Provided are systems and methods for contrastive learning of visual representations. In particular, the present disclosure provides systems and methods that leverage particular data augmentation schemes and a learnable nonlinear transformation between the representation and the contrastive loss to provide improved visual representations. In contrast to certain existing techniques, the contrastive self-supervised learning algorithms described herein do not require specialized architectures or a memory bank. Some example implementations of the proposed approaches can be referred to as a simple framework for contrastive learning of representations or “SimCLR.” Further example aspects are described below and provide the following benefits and insights.Type: GrantFiled: April 13, 2020Date of Patent: June 7, 2022Assignee: GOOGLE LLCInventors: Ting Chen, Simon Kornblith, Mohammad Norouzi, Geoffrey Everest Hinton
-
Patent number: 11354786Abstract: The invention relates to a method for masking an image (20) of an image sequence with a mask (30), wherein a mask colour is determined according to an image colour of the image (20) in the environment of the mask (30), and a colour of the mask (30) is set according to the determined mask colour or with the determined mask colour. The invention also relates to a computer program, a machine-readable storage medium and an electronic storage unit.Type: GrantFiled: October 10, 2017Date of Patent: June 7, 2022Assignee: Robert Bosch GmbHInventors: Bart Corstiaans, Konrad Simon, Rainer Bauereiss, Sachin Khanna
-
Patent number: 11348211Abstract: An embodiment of the present invention provides an image processing method based on a convolution internet algorithm, a device, an electronic apparatus and a computer readable storage medium thereof. The method includes: determining a padding radius for each convolution calculation layer; summing the padding radiuses; acquiring an input image of a convolution internet algorithm; padding the input image to acquire a padded image according to the sum; determining a calculation range for each convolution calculation layer; implementing a convolution internet algorithm calculation to the padded image according to each convolution calculation layer; acquiring an output image. Image processing device, electronic apparatus, and computer readable storage medium of a convolution internet algorithm are provided.Type: GrantFiled: December 27, 2019Date of Patent: May 31, 2022Assignee: TCL China Star Optoelectronics Technology Co., Ltd.Inventor: Chinghong Lai
-
Patent number: 11342943Abstract: Techniques to improve storage, transmission and security of data are included. One or more methods, apparatuses, and articles of manufacture employ one or more color-channels, ultraviolet layers, infrared layers, and/or luminance layers to encode data on or along a physical medium, where the encoding includes utilizing one or more of those layers to encode an error-correcting code (ECC), such as a Hamming code with the data.Type: GrantFiled: June 9, 2020Date of Patent: May 24, 2022Assignee: Capital One Services, LLCInventors: Austin Grant Walters, Jeremy Edward Goodsitt
-
Patent number: 11335118Abstract: A signal output unit outputs an acquired signal. A signal accumulation unit accumulates signals. A signal attribute value display unit displays a value of an attribute related to an element constituting a target represented by the acquired signal or a signal generation source in a state in which an instruction for changing the value of the attribute is able to be received. A changed attribute value acquisition unit acquires a changed value of the attribute when the instruction for changing the value of the attribute is received. A signal retrieval unit retrieves, from the signals accumulated in the signal accumulation unit, a signal similar to a changed signal when the value of the attribute has been changed on the basis of the changed value of the attribute acquired by the changed attribute value acquisition unit.Type: GrantFiled: May 1, 2018Date of Patent: May 17, 2022Assignee: NIPPON TELEGRAPH AND TELEPHONE CORPORATIONInventors: Takuhiro Kaneko, Kaoru Hiramatsu, Kunio Kashino
-
Patent number: 11335081Abstract: A method for automatic extraction of data from a graph, including text area locating and text box classification; locating of coordinate axes, and locating of the positions of hatch marks on the coordinate axes; legend locating and information extraction; extracting corresponding bar or polyline connected components according to legend color, and filtering and classification; determining key points on the X-axis and locating a corresponding X-axis label for each key point; locating key points of the bars and polyline according to the X-axis key points, determining labeled numerical text boxes that correspond to the key points, and identifying the numerical text; calculating a corresponding value for each pixel, and estimating corresponding values of the key points of the bars or polyline; determining a final result according to a difference between the estimated values and the recognized labeled values.Type: GrantFiled: August 19, 2020Date of Patent: May 17, 2022Assignee: HANGZHOU DIANZI UNIVERSITYInventors: Jinliang Yao, Yuelun Yu, Chuang Hu
-
Patent number: 11328453Abstract: Proposed in various embodiments of the present disclosure are a device and a method for image processing, whereby image conversion by dilation is carried out in a multimedia system. To this end, an electronic device for image processing may identify between a first unused area in a first picture and a second unused area in a second picture, the first unused area and the second unused area having the same time information. The electronic device may acquire fill values for replacing the first unused area and the second unused area, on the basis of a value acquired from the first unused area and a value acquired from the second unused area. The electronic device may acquire a first converted picture and a second converted picture by replacing the first unused area and the second unused area by using the acquired fill values.Type: GrantFiled: April 11, 2019Date of Patent: May 10, 2022Assignee: Samsung Electronics Co., Ltd.Inventors: Youngho Oh, Sungryeul Rhyu, Jihwan Woo
-
Patent number: 11328448Abstract: According to one embodiment, an image processing method causes a computer to function as an image processing apparatus including an acquisition unit and a control unit. The acquisition unit acquires image data. The control unit performs a control process of detecting a character string of each color included in the image data and associating the character string with the color using the character string of the color.Type: GrantFiled: September 19, 2019Date of Patent: May 10, 2022Assignee: TOSHIBA TEC KABUSHIKI KAISHAInventor: Masatoshi Shinohara