Patents Examined by Leon Flores
-
Patent number: 11200451Abstract: A template determining apparatus including an attribute distribution determination unit configured to determine a distribution of a specific attribute in a plurality of images; and a template determination unit configured to adaptatively determine a template set from the plurality of images according to the determined distribution of the specific attribute of the plurality of images. Where the determined template set will be used for image normalization.Type: GrantFiled: November 11, 2019Date of Patent: December 14, 2021Assignee: Canon Kabushiki KaishaInventors: Yaohai Huang, Jianteng Peng, Weihong Deng, Jiani Hu
-
Patent number: 11195051Abstract: The invention relates to a method for person re-identification based on deep model with multi-loss fusion training strategy. The method uses a deep learning technology to perform preprocessing operations such as flipping, clipping, random erasing and style transfer, and then feature extraction is performed through a backbone network model; joint training of a network is performed by fusing a plurality of loss functions. Compared with other deep learning-based person re-identification algorithms, the present invention greatly improves the performance of person re-identification by adopting a plurality of preprocessing modes, the fusion of three loss functions and effective training strategy.Type: GrantFiled: March 9, 2020Date of Patent: December 7, 2021Assignees: Tongji University, Hefei University of Technology, Beijing E-Hualu Info Technology Co., Ltd.Inventors: Deshuang Huang, Sijia Zheng, Zhongqiu Zhao, Xinyong Zhao, Jianhong Sun, Yang Zhao, Yongjun Lin
-
Patent number: 11189014Abstract: A method for processing an image includes: an image to be processed with a first resolution is acquired; and the image to be processed is processed by a target neural network model to obtain a target image, the target image being a denoised image with a second resolution, the second resolution being higher than the first resolution, and the target neural network model including a first preset number of convolutional layers and a second preset number of sub-pixel up-sampling portions.Type: GrantFiled: November 18, 2019Date of Patent: November 30, 2021Assignee: BEIJING XIAOMI MOBILE SOFTWARE CO., LTD.Inventors: Hailong Ma, Xiangxiang Chu, Qike Zhao
-
Patent number: 11182923Abstract: A method, a computing device, and a system for monitoring postures are proposed. The method includes the following steps. An image sequence captured on a monitored area including a monitored subject is received. A first stable value corresponding to a first stage of stability of the monitored subject is detected, where the first stable value is associated with a position corresponding to a posture of the monitored subject remaining consecutively stable within a first predetermined time period. A second stable value corresponding to a second stage of stability of the monitored subject is detected, where the second stable value is associated with a position corresponding to a posture of the monitored subject remaining consecutively stable within a second predetermined time period. The first stable value and the second stable value are compared to accordingly determine whether a posture of the monitored subject has changed.Type: GrantFiled: July 11, 2019Date of Patent: November 23, 2021Assignee: Wistron CorporationInventors: Ting-Feng Ju, Chan-Hsuan Yu, Kuo-Hsien Lu
-
Patent number: 11182619Abstract: For point-of-interest determination and display, a processor detects an image event during a videoconference. The processor determines a point-of-interest for the video image of the videoconference based on the image event. The video image is at least a 180-degree image and the point-of-interest is a portion of the video image. The processor displays the point-of-interest from the video image on a display.Type: GrantFiled: January 14, 2020Date of Patent: November 23, 2021Assignee: LENOVO (Singapore) PTE. LTD.Inventors: Adam Jerome Cavenaugh, David W. Douglas, Kazuo Fujii, Kenneth Seethaler
-
Patent number: 11176823Abstract: A computer includes a processor and a memory, the memory storing instructions executable by the processor to input an image to a first layer of a machine learning program, the first layer trained to identify one or more quadrilateral regions in the image, upon identifying the one or more quadrilateral regions, input the collected image to a second layer of a machine learning program, the second layer trained to identify a plurality of sets of vertices, each set of vertices defining a respective polygonal area, identify one of the polygonal areas in which to park a vehicle, and actuate one or more vehicle components to move the vehicle into the identified polygonal area.Type: GrantFiled: March 30, 2020Date of Patent: November 16, 2021Assignee: FORD GLOBAL TECHNOLOGIES, LLCInventors: Mayar Arafa, Vidya Nariyambut murali, Xianling Zhang, Nikita Jaipuria, Rohan Bhasin
-
Patent number: 11176658Abstract: A method comprising determining a binary classification value for each of a plurality of data instances based on a first threshold value assigned to each of the plurality of data instances; applying at least one clustering model to a first subset of the plurality of data instances to identify one or more dominant clusters of data instances; determining a second threshold value to assign to a second plurality of data instances that are included within the one or more dominant clusters of data instances; and redetermining a binary classification value for each of the plurality of data instances based on the second threshold value assigned to the second plurality of data instances and the first threshold value, wherein the first threshold value is assigned to at least a portion of data instances of the plurality of data instances that are not included in the second plurality of data instances.Type: GrantFiled: September 16, 2019Date of Patent: November 16, 2021Assignee: Intel CorporationInventors: Bikram Baidya, Allan Gu, Vivek K. Singh, Kumara Sastry, Abde Ali Hunaid Kagalwalla
-
Patent number: 11170258Abstract: A method of training a neural network for reducing noise in an image comprises obtaining a plurality of input images from an imaging device. The method comprises generating a plurality of target images by combining a subset of the input images that each depict a same object, to reduce a noise of the target image. The method comprises generating a plurality of training pairs, wherein a training pair comprises one of the target images and a training image based on at least one but not all of the input images of the subset of input images corresponding to the one of the target image. The method comprises training a neural network using the plurality of training pairs.Type: GrantFiled: May 16, 2018Date of Patent: November 9, 2021Assignee: RetinAl Medical AGInventors: Carlos Ciller Ruiz, Sandro de Zanet, Stefanos Apostolopoulos
-
Patent number: 11170260Abstract: A system for determining the importance of encoded image components for artificial intelligence tasks includes an image capture or storage unit, a processor and a communication interface. The processor can receive components of transformed domain image data from the one or more image capture or storage units across the communication interface. The processor can be configured to determine the relative importance of the components of the transformed domain image data for an artificial intelligence task.Type: GrantFiled: November 14, 2019Date of Patent: November 9, 2021Assignee: Alibaba Group Holding LimitedInventors: Kai Xu, Minghai Qin, Yuhao Wang, Fei Sun, Yen-kuang Chen, Yuan Xie
-
Patent number: 11157724Abstract: An image recognition device executes a Hilbert scan of frame image data constituting moving-image data to generate one-dimensional spatial image data, and further arrays the one-dimensional spatial image data in a time direction to generate two-dimensional spatio-temporal image data that holds spatial information and temporal information. The image recognition device converts the moving-image data into the two-dimensional spatio-temporal image data while holding the spatial and temporal information. By means of a CNN unit, the image recognition device executes a convolution process wherein a two-dimensional filter is used on the spatio-temporal image data to image-recognize a behavior of a pedestrian who is a recognition object.Type: GrantFiled: July 31, 2018Date of Patent: October 26, 2021Assignees: EQUOS RESEARCH CO., LTD., KYUSHU INSTITUTE OF TECHNOLOGYInventors: Hideo Yamada, Ryuya Muramatsu, Masatoshi Shibata, Shuichi Enokida
-
Patent number: 11151359Abstract: The present disclosure provides a face swap method, a face swap device, a host terminal and an audience terminal. The method includes: starting a face swap prompt; recognizing and determining a first face image in a first preset image of the host terminal according to the face swap prompt; receiving a second face image sent by the audience terminal; and replacing the first face image with the second face image to obtain a first face-swapped image.Type: GrantFiled: May 21, 2019Date of Patent: October 19, 2021Assignee: JOYME PTE. LTD.Inventors: Wenpei Hou, He Li, Chenying Wang, Diqin Jiao, Huan Long
-
Patent number: 11151728Abstract: The present invention discloses a structure monitoring system comprising a plurality of marking units disposed on a structural object, a monitoring device monitoring the plurality of marking units remotely, and a data processing device connected to the monitoring device. The aforementioned data processing device further comprises a data receiving module and a data processing module. The data receiving module receives a data detected by the monitoring device, and the data processing module calculates and analyzes the information. In addition, a structural monitoring method has also been provided.Type: GrantFiled: December 20, 2019Date of Patent: October 19, 2021Assignee: SHIP AND OCEAN INDUSTRIES R&D CENTERInventors: Cheng-Hsien Chung, Hua-Tung Wu, Hsin-Haou Huang
-
Patent number: 11151692Abstract: A method for processing an image includes: an image to be processed with a first resolution is acquired; and the image to be processed is processed by a target neural network model to obtain a target image, the target image being a denoised image with a second resolution, the second resolution being higher than the first resolution, and the target neural network model including a first preset number of convolutional layers and a second preset number of sub-pixel up-sampling portions.Type: GrantFiled: November 18, 2019Date of Patent: October 19, 2021Assignee: BEIJING XIAOMI MOBILE SOFTWARE CO., LTD.Inventors: Hailong Ma, Xiangxiang Chu, Qike Zhao
-
Patent number: 11138452Abstract: A computer, including a processor and a memory, the memory including instructions to be executed by the processor to generate two or more stereo pairs of synthetic images and generate two or more stereo pairs of real images based on the two or more stereo pairs of synthetic images using a generative adversarial network (GAN), wherein the GAN is trained using a six-axis degree of freedom (DoF) pose determined based on the two or more pairs of real images. The instructions can further include instructions to train a deep neural network based on a sequence of real images and operate a vehicle using the deep neural network to process a sequence of video images acquired by a vehicle sensor.Type: GrantFiled: October 8, 2019Date of Patent: October 5, 2021Assignee: FORD GLOBAL TECHNOLOGIES, LLCInventors: Punarjay Chakravarty, Praveen Narayanan, Nikita Jaipuria, Gaurav Pandey
-
Patent number: 11138419Abstract: A learning device generates a plurality of learning images in which a distance image representing a distance from a reference position to each position of a human body or each position of an object and a part image for identifying each part of the human body or a part of the object are associated with each other. The learning device corrects, based on a distance image and a part image of the learning image, a value of a region corresponding to a part of the object among regions of the distance image. The learning device learns, based on a plurality of learning images including a corrected distance image, an identifier in which characteristics of the distance image and a part of the human body or a part of the object are associated with each other.Type: GrantFiled: October 15, 2019Date of Patent: October 5, 2021Assignee: FUJITSU LIMITEDInventors: Shoichi Masui, Hiroaki Fujimoto, Kazuhiro Yoshimura, Takuya Sato, Kazuo Sasaki
-
Patent number: 11132778Abstract: The present invention reduces the amount of data outputted while maintaining the accuracy of an analysis process with a small delay amount, by an image analysis apparatus provided with: a deduction unit that deduces a second quality concerning an object in a second image, which is different from a first image associated with object data relating to an object to be inputted, on the basis of a first quality concerning the object relating to the object data and on the basis of the state of the object in the second image, the state being obtained by using a state model for deducing the position and the size of the object from the object data, while using a quality model for deducing the second quality concerning the object; and a determination unit that determines whether to use the object data for analysis on the basis of the deduced second quality.Type: GrantFiled: March 29, 2017Date of Patent: September 28, 2021Assignee: NEC CORPORATIONInventor: Yosuke Iwamatsu
-
Patent number: 11120308Abstract: A vehicle damage detection method based on image analysis, an electronic device, and a storage medium are provided. In the vehicle damage detection method, query images are obtained by filtering received images through a pre-trained Single Shot MultiBox Detector (SSD) object detection model, and a feature vector of each of the query images is obtained by inputting each of the query images into a residual network. Target output data is obtained using a Transformer model, similar images of the query images are obtained by processing the target output data using a pre-trained similarity judgment model. Loss of a current vehicle damage assessment case is evaluated based on similar cases, and evaluated loss is outputted. By utilizing the vehicle damage detection method, effectiveness of the vehicle damage detection is improved, and automatic evaluation of a loss is achieved.Type: GrantFiled: December 24, 2019Date of Patent: September 14, 2021Assignee: Ping An Technology (Shenzhen) Co., Ltd.Inventors: Kun Li, Hao Zhang, Ruei-Sung Lin, Mei Han
-
Patent number: 11120533Abstract: An information processing method includes: acquiring fixed values which are imaging conditions of a first image formed in a state in which a background and a predetermined subject located in front of the background are within an imaging visual field, from the first image; acquiring a learning image as background learning data, wherein the learning image is captured using the fixed values in a second image formed in a state in which the subject is outside of the imaging visual field; extracting a difference image between the background learning data and an input image obtained by capturing the first image using the fixed values; and generating a combined image obtained by combining the difference image with a combination background image.Type: GrantFiled: July 11, 2019Date of Patent: September 14, 2021Assignee: Roland CorporationInventor: Atsushi Kida
-
Patent number: 11120275Abstract: The present disclosure provides a visual perception method, an apparatus, a device and a medium based on an autonomous vehicle, the method includes inputting an obtained first visual perception image collected by the autonomous vehicle into a first neural network model, recognizing multi-channel feature information of at least one target recognition object to be recognized, to eliminate redundant feature information in the first visual perception image; further, inputting the multi-channel feature information of the at least one target recognition object to be recognized into at least one sub-neural network model in a second neural network model respectively, to obtain at least one target recognition object; where there is a one to one correspondence between the target recognition object and the sub-neural network model.Type: GrantFiled: July 18, 2019Date of Patent: September 14, 2021Assignee: Baidu Online Network Technology (Beijing) Co., Ltd.Inventors: Jiajia Chen, Ji Wan, Tian Xia
-
Patent number: 11093753Abstract: A visual SLAM system comprises a plurality of keyframes including a keyframe, a current keyframe, and a previous keyframe, a dual dense visual odometry configured to provide a pairwise transformation estimate between two of the plurality of keyframes, a frame generator configured to create keyframe graph, a loop constraint evaluator adds a constraint to the receiving keyframe graph, and a graph optimizer configured to produce a map with trajectory.Type: GrantFiled: June 26, 2017Date of Patent: August 17, 2021Assignee: Robert Bosch GmbHInventors: Soohwan Kim, Benzun Pious Wisely Babu, Zhixin Yan, Liu Ren