Patents by Inventor Yihao Zhang
Yihao Zhang has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Publication number: 20240386623Abstract: Embodiments described herein provide a method of image generation. The method includes a fixed diffusion model, and a trainable diffusion model. The fixed diffusion model may be pretrained on a large training corpus. The trainable diffusion model may be used to control the image generation of the fixed diffusion model by modifying internal representations of the fixed diffusion model. A task instruction may be provided in addition to a text prompt, and the task instruction may guide the trainable diffusion model together with the visual conditions. The visual conditions may be adapted according to the task instruction. During training, a fixed number of task instructions may be used. At inference, unseen task instructions may be used by combining convolutional kernels of the visual condition adapter.Type: ApplicationFiled: September 29, 2023Publication date: November 21, 2024Inventors: Ning YU, Can QIN, Shu ZHANG, Yihao FENG, Xinyi YANG, Ran XU
-
Publication number: 20240343277Abstract: A metro rail corrugation measurement method based the Seq2Seq model and vibration and noise data fusion comprises the following steps: Constructing a time series data set corresponding to floor longitudinal acceleration and train speed based on historical data of metro trains; Constructing the metro train mileage matching prediction model based on the Seq2Seq model, training the metro train mileage matching prediction model by time series data sets, and inputting the inside floor longitudinal acceleration of the metro train to be detected into the trained metro train mileage matching prediction model to obtain the running speed of the metro train to be detected. According to the vibration and noise data of the metro train to be detected, the vibration and noise composite index of the rail corrugation is calculated to determine the wavelength and amplitude information corresponding to the rail corrugation.Type: ApplicationFiled: August 24, 2023Publication date: October 17, 2024Applicant: BEIJING JIAOTONG UNIVERSITYInventors: Hong XIAO, Yang WANG, Liang GAO, Shuwei FANG, Xiubo LIU, Zhihai ZHANG, Guangming SHI, Feng JIN, Gang WANG, Yan XIAO, Libin YE, Yihao CHI, Guangpeng LIU, Shaolei WEI, Guangsheng CHEN, Zhongxia QIAN, Jianjun MA, Chang XIAO, Yuze CAO, Yawen ZHANG
-
Publication number: 20240345999Abstract: This application discloses a database system update method, applied to a database including a first node, a second node, and a shared storage device. The method includes: The first node sends a first instruction to the second node, where the first instruction is used to request storage location information of first data in the shared storage device, and the first data is data transferred by the second node to the first node for processing; and the first node obtains first information sent by the second node, and obtains mapping information from the shared storage device based on the first information, where the mapping information indicates a storage location of the first data in the shared storage device. The method is used to improve update efficiency of the database system, and ensure that different nodes can simultaneously process different data in a shared storage device.Type: ApplicationFiled: June 27, 2024Publication date: October 17, 2024Inventors: Qiong ZHANG, Wei CHENG, Biao WANG, Hanchen ZHOU, Xuli LI, Lening CHEN, Lin LI, Chao MA, Yihao SONG
-
Publication number: 20240303882Abstract: Embodiments described herein provide a feedback based instructional image editing framework that employs a diffusion process to follow user instruction for image editing. A diffusion model is fine-tuned using a reward model, which may be trained via human annotation. The training of the reward model may be done by having the image editing model output a number of images, which a human annotator ranks based on their alignment with the original image and a given instruction.Type: ApplicationFiled: July 12, 2023Publication date: September 12, 2024Inventors: Shu Zhang, Xinyi Yang, Yihao Feng, Ran Xu, Ning Yu, Chia-Chih Chen
-
Publication number: 20240100208Abstract: An innovative equipment system comprises a tap water reservoir system, a disinfectant reactor system, a pump system, a disinfection and safety door spraying system, a sewage recycling system and a sewage treatment system, wherein the tap water reservoir system is mainly configured to store tap water in advance according to actual cargo volumes and disinfection requirements, and generally in tons; and the disinfectant reactor system comprises boron-doped diamond electrode modules, a reactor indicating lamp module, a visible reactor, a smart control screen, at least one water tank and a structure protection module. The Innovative disinfection and sterilization equipment system can easily kill bacteria and viruses such as Escherichia coli, Staphylococcus aureus, Salmonella typhimurium, and new coronavirus in a short time and enhance disinfection efficiency.Type: ApplicationFiled: November 7, 2022Publication date: March 28, 2024Inventors: Pengzhou LEI, Yihao ZHANG, Lingli SUN, Junhong CHEN, Yingqiang ZHAO, Bo WANG, Xiangye ZENG, Zhanjun ZHANG, Jiandong ZHANG, Zhiqiang SHI
-
Publication number: 20220398531Abstract: A method for automatically creating an equipment inventory from a plurality of images of a site. The method includes receiving, by a processor, the plurality of images of the site, labeling, by the processor, equipment present in each image of the plurality of images of the site, creating, by the processor, an overlap curve for neighboring images of the plurality of images of the site, determining, by the processor, a useful image of each side of the site based on the overlap curve, the useful image including a subset of the plurality of images, creating, by the processor, an overlay of equipment in the useful image of each side of the site, counting, by the processor, equipment present in the overlay of each side of the site, and generating, by the processor, the equipment inventory by adding a number of equipment counted in the overlay of each side of the site.Type: ApplicationFiled: August 18, 2022Publication date: December 15, 2022Inventors: Gordon WILFONG, Yihao ZHANG, Iraj SANIEE, Maria Laura MAAG
-
Publication number: 20220318631Abstract: Systems, methods, and software of processing Deep Neural Networks (DNN) comprising a plurality of nodes with connections between the nodes. In one embodiment, a system initializes the DNN by assigning initial values to weights associated with the connections, and identifies an initial subset of the connections randomly selected from the DNN. The initial subset after training with a training dataset has a training accuracy below an accuracy threshold. The system performs a growth process by adding additional connections from the DNN to the initial subset to generate a qualified subset of the connections. The qualified subset of the connections after training with the training dataset has a training accuracy that reaches the accuracy threshold.Type: ApplicationFiled: April 5, 2021Publication date: October 6, 2022Inventors: Iraj Saniee, Yihao Zhang, Bradley Magnetta
-
Patent number: 11238274Abstract: An image feature extraction method for person re-identification includes performing person re-identification by means of aligned local descriptor extraction and graded global feature extraction; performing the aligned local descriptor extraction by processing an original image by affine transformation and performing a summation pooling operation on image block features of same regions to obtain an aligned local descriptor; reserving spatial information between inner blocks of the image for the aligned local descriptor; and performing the graded global feature extraction by grading a positioned pedestrian region block and solving a corresponding feature mean value to obtain a global feature. The method can resolve the problem of feature misalignment caused by posture changes of pedestrian, etc., and eliminate the effect of unrelated backgrounds on re-recognition, thus improving the precision and robustness of person re-identification.Type: GrantFiled: December 27, 2017Date of Patent: February 1, 2022Assignee: Peking UniversityInventors: Wenmin Wang, Yihao Zhang, Ronggang Wang, Ge Li, Shengfu Dong, Zhenyu Wang, Ying Li, Wen Gao
-
Patent number: 11232115Abstract: A computer system may use machine learning to determine functional zones within a geographical region. The computer system may determine position data and functional data for locations within a geographical region. The computer system may use machine learning to determine functional zones with a two-stage clustering approach. During the first stage, the computer system may determine geoclusters by clustering locations based on the position data. During the second stage, the computer system may determine functional zones by clustering the geoclusters based on the functional data.Type: GrantFiled: April 11, 2018Date of Patent: January 25, 2022Assignee: Nokia Technologies OyInventors: Jin Cao, Ran He, Yihao Zhang, Jingyuan Yang
-
Patent number: 11100370Abstract: Disclosed is a deep discriminative network for person re-identification in an image or a video. Concatenation are carried out on different input images on a color channel by constructing a deep discriminative network, and an obtained splicing result is defined as an original difference space of different images. The original difference space is sent into a convolutional network. The network outputs the similarity between two input images by learning difference information in the original difference space, thereby realizing person re-identification. The features of an individual image are not learnt, and concatenation are carried out on input images on a color channel at the beginning, and difference information is learnt on an original space of the images by using a designed network. By introducing an Inception module and embedding the same into a model, the learning ability of a network can be improved, and a better differentiation effect can be achieved.Type: GrantFiled: January 23, 2018Date of Patent: August 24, 2021Assignee: Peking University Shenzhen Graduate SchoolInventors: Wenmin Wang, Yihao Zhang, Ronggang Wang, Ge Li, Shengfu Dong, Zhenyu Wang, Ying Li, Hui Zhao, Wen Gao
-
Publication number: 20210150194Abstract: An image feature extraction method for person re-identification includes performing person re-identification by means of aligned local descriptor extraction and graded global feature extraction; performing the aligned local descriptor extraction by processing an original image by affine transformation and performing a summation pooling operation on image block features of same regions to obtain an aligned local descriptor; reserving spatial information between inner blocks of the image for the aligned local descriptor; and performing the graded global feature extraction by grading a positioned pedestrian region block and solving a corresponding feature mean value to obtain a global feature. The method can resolve the problem of feature misalignment caused by posture changes of pedestrian, etc., and eliminate the effect of unrelated backgrounds on re-recognition, thus improving the precision and robustness of person re-identification.Type: ApplicationFiled: December 27, 2017Publication date: May 20, 2021Inventors: Wenmin Wang, Yihao Zhang, Ronggang Wang, Ge Li, Shengfu Dong, Zhenyu Wang, Ying Li, Wen Gao
-
Publication number: 20210150268Abstract: Disclosed is a deep discriminative network for person re-identification in an image or a video. Concatenation are carried out on different input images on a color channel by constructing a deep discriminative network, and an obtained splicing result is defined as an original difference space of different images. The original difference space is sent into a convolutional network. The network outputs the similarity between two input images by learning difference information in the original difference space, thereby realizing person re-identification. The features of an individual image are not learnt, and concatenation are carried out on input images on a color channel at the beginning, and difference information is learnt on an original space of the images by using a designed network. By introducing an Inception module and embedding the same into a model, the learning ability of a network can be improved, and a better differentiation effect can be achieved.Type: ApplicationFiled: January 23, 2018Publication date: May 20, 2021Inventors: Wenmin Wang, Yihao Zhang, Ronggang Wang, Ge Li, Shengfu Dong, Zhenyu Wang, Ying Li, Hui Zhao, Wen Gao
-
Publication number: 20200226188Abstract: System, method, and software for recommending documents to users. A recommendation system identifies recent documents and historical information indicating consumption by users of past documents. The system generates a historical reading matrix based on the historical information, and generates a vector matrix for the recent documents and the past documents. The system generates an estimated reading matrix based on the historical reading matrix and the vector matrix, with the estimated reading matrix having first entries that represent actual reading scores for the past documents. The system calculates estimated reading scores for the recent documents based on the vector matrix, and populates second entries of the estimated reading matrix corresponding to the recent documents with the estimated reading scores. The system performs a factorization on the estimated reading matrix to generate a refined reading matrix, and generates recommendations of the recent documents based on the refined reading matrix.Type: ApplicationFiled: January 14, 2019Publication date: July 16, 2020Inventors: William Kennedy, Gordon Wilfong, Yihao Zhang, Hung Nguyen
-
Publication number: 20190318028Abstract: A computer system may use machine learning to determine functional zones within a geographical region. The computer system may determine position data and functional data for locations within a geographical region. The computer system may use machine learning to determine functional zones with a two-stage clustering approach. During the first stage, the computer system may determine geoclusters by clustering locations based on the position data. During the second stage, the computer system may determine functional zones by clustering the geoclusters based on the functional data.Type: ApplicationFiled: April 11, 2018Publication date: October 17, 2019Inventors: Jin Cao, Ran He, Yihao Zhang, Jingyuan Yang
-
Patent number: 9760298Abstract: Techniques are provided for anonymizing data in a data stream.Type: GrantFiled: June 22, 2015Date of Patent: September 12, 2017Assignee: Alcatel-Lucent USA Inc.Inventors: Daniel M. Andrews, Avinash Vyas, Gordon T. Wilfong, Yihao Zhang
-
Patent number: 9667499Abstract: A capability for sparsifying a representation of pairwise cost information is presented herein. The capability for sparsifying a representation of pairwise cost information may be used to sparsify a representation of pairwise cost information for a set of nodes. The sparsification of a representation of pairwise cost information for a set of nodes may provide thereby a sparsified representation of the pairwise cost information for the set of nodes. The sparsification of the representation of pairwise cost information for the set of nodes may be based on clustering of the nodes of the set of nodes into clusters. The sparsification of the representation of pairwise cost information for the set of nodes may be based on calculation of intra-cluster costs and inter-cluster costs, where the intra-cluster costs and inter-cluster costs are calculated based on the pairwise cost information of the representation of the pairwise cost information for the set of nodes.Type: GrantFiled: September 12, 2014Date of Patent: May 30, 2017Assignees: Alcatel Lucent, Alcatel-Lucent USA Inc.Inventors: Yihao Zhang, Gordon Wilfong, Michael Scharf
-
Patent number: 9507932Abstract: A capability for providing policy enforcement in a topology abstraction system is presented. The capability for providing policy enforcement in a topology abstraction system may support use of topology abstraction policies to control abstraction of topology information of a topology (e.g., a network topology of a communication network or any other suitable type of topology). The capability for providing policy enforcement in a topology abstraction system providing an abstract representation of a topology may support use of topology abstraction policies to control selection (or acceptance) of topology elements for inclusion within the abstract representation of a topology and filtering (or rejection) of topology elements from being included within the abstract representation of a topology.Type: GrantFiled: September 12, 2014Date of Patent: November 29, 2016Assignee: Alcatel LucentInventors: Michael Scharf, Thomas Voith, Manuel Stein, Yihao Zhang, Gordon Wilfong
-
Publication number: 20160313939Abstract: Techniques are provided for anonymizing data in a data stream.Type: ApplicationFiled: June 22, 2015Publication date: October 27, 2016Inventors: Daniel M. Andrews, Avinash Vyas, Gordon T. Wilfong, Yihao Zhang
-
Patent number: D951902Type: GrantFiled: January 21, 2020Date of Patent: May 17, 2022Assignee: Shenzhen Chuangwei-RGB Electronics Co., LtdInventors: Yihao Zhang, Xiangcheng Liu, Shuxiao Wei, Qiang Li, Linrong Feng, Xingkang Liu
-
Patent number: D1017563Type: GrantFiled: May 27, 2021Date of Patent: March 12, 2024Assignee: Shenzhen Chuangwei-RGB Electronics Co., Ltd.Inventors: Qiang Li, Xiangcheng Liu, Bosheng Tang, Yihao Zhang