Patents by Inventor Ziwei Liu
Ziwei Liu 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).
-
Patent number: 11301719Abstract: A semantic segmentation model training method includes: performing, by a semantic segmentation model, image semantic segmentation on at least one unlabeled image to obtain a preliminary semantic segmentation result as the category of the unlabeled image; obtaining, by a convolutional neural network based on the category of the at least one unlabeled image and the category of at least one labeled image, sub-images respectively corresponding to the at least two images and features corresponding to the sub-images, where the at least two images comprise the at least one unlabeled image and the at least one labeled image, and the at least two sub-images carry the categories of the corresponding images; and training the semantic segmentation model on the basis of the categories of the at least two sub-images and feature distances between the at least two sub-images.Type: GrantFiled: December 25, 2019Date of Patent: April 12, 2022Assignee: BEIJING SENSETIME TECHNOLOGY DEVELOPMENT CO., LTD.Inventors: Xiaohang Zhan, Ziwei Liu, Ping Luo, Chen Change Loy, Xiaoou Tang
-
Patent number: 11222211Abstract: A method and an apparatus for segmenting a video object, an electronic device, a storage medium, and a program include: performing, among at least some frames of a video, inter-frame transfer of an object segmentation result of a reference frame in sequence from the reference frame, to obtain an object segmentation result of at least one other frame among the at least some frames; determining other frames having lost objects with respect to the object segmentation result of the reference frame among the at least some frames; using the determined other frames as target frames to segment the lost objects, so as to update the object segmentation results of the target frames; and transferring the updated object segmentation results of the target frames to the at least one other frame in the video in sequence. The accuracy of video object segmentation results can therefore be improved.Type: GrantFiled: December 29, 2018Date of Patent: January 11, 2022Assignee: BEIJING SENSETIME TECHNOLOGY DEVELOPMENT CO., LTDInventors: Xiaoxiao Li, Yuankai Qi, Zhe Wang, Kai Chen, Ziwei Liu, Jianping Shi, Ping Luo, Chen Change Loy, Xiaoou Tang
-
Publication number: 20210383154Abstract: An image processing method includes: a color feature extracted from a first image is acquired; a customized mask feature is acquired, the customized mask feature being configured to indicate a regional position of the color feature in the first image; and the color feature and the customized mask feature are input to a feature mapping network to perform image attribute edition to obtain a second image.Type: ApplicationFiled: August 22, 2021Publication date: December 9, 2021Applicant: SHENZHEN SENSETIME TECHNOLOGY CO., LTD.Inventors: Chenghan LI, Ziwei LIU, Lingyun WU, Ping LUO
-
Publication number: 20210279892Abstract: An image processing method and a device, and a network training method and a device are provided. The image processing method includes determining a guide group arranged on an image to be processed and directed at a target object, the guide group comprising at least one guide point, and the guide point being used to indicate the position of a sampling pixel, and the magnitude and direction of the motion speed of the sampling pixel; and on the basis of the guide point in the guide group and the image to be processed, performing optical flow prediction to obtain the motion of the target object in the image to be processed.Type: ApplicationFiled: May 25, 2021Publication date: September 9, 2021Inventors: Xiaohang ZHAN, Xingang PAN, Ziwei LIU, Dahua LIN, Chen Change LOY
-
Patent number: 11073670Abstract: A sealed multiport splitter device and method are disclosed. The fiber optic multiport comprises housing with an interior defining a first chamber and an adjacent second chamber. An optical splitter with a plurality of splitter legs connect to a plurality of optical fibers located in the first chamber. Potting material is disposed in the first chamber to physically secure the optical splitter, the splitter legs and the optical fibers. A wall having a first face and a second face separates the first chamber from the second chamber. The wall has a plurality of slots extended from the first face to the second face. The plurality of optical fibers route through the plurality of slots between the first chamber and the second chamber. A blocking material is adjacent to the wall to inhibit ingress of the potting material into the second chamber while the potting material is being cured.Type: GrantFiled: August 12, 2016Date of Patent: July 27, 2021Assignee: Corning Optical Communications LLCInventors: Monique Lise Cote, Troy Dean Lindblad, Ziwei Liu
-
Patent number: 11049217Abstract: An example of the present disclosure provides methods, apparatuses and devices for magnifying a feature map, and a computer readable storage medium. The method includes: receiving a source feature map to be magnified; obtaining N reassembly kernels corresponding to each source position in the source feature map by performing convolution on the source feature map, wherein N refers to a square of a magnification factor of the source feature map; obtaining, for each of the reassembly kernels, a normalized reassembly kernel by performing normalization; obtaining, for each source position in the source feature map, N reassembly features corresponding to the source position by reassembling features of a reassembly region determined according to the source position with N normalized reassembly kernels corresponding to the source position; and generating a target feature map according to the N reassembly features corresponding to each source position in the source feature map.Type: GrantFiled: December 15, 2020Date of Patent: June 29, 2021Assignee: Beijing Sensetime Technology Development Co., Ltd.Inventors: Jiaqi Wang, Kai Chen, Rui Xu, Ziwei Liu, Chen Change Loy, Dahua Lin
-
Publication number: 20210104015Abstract: An example of the present disclosure provides methods, apparatuses and devices for magnifying a feature map, and a computer readable storage medium. The method includes: receiving a source feature map to be magnified; obtaining N reassembly kernels corresponding to each source position in the source feature map by performing convolution on the source feature map, wherein N refers to a square of a magnification factor of the source feature map; obtaining, for each of the reassembly kernels, a normalized reassembly kernel by performing normalization; obtaining, for each source position in the source feature map, N reassembly features corresponding to the source position by reassembling features of a reassembly region determined according to the source position with N normalized reassembly kernels corresponding to the source position; and generating a target feature map according to the N reassembly features corresponding to each source position in the source feature map.Type: ApplicationFiled: December 15, 2020Publication date: April 8, 2021Inventors: Jiaqi WANG, Kai CHEN, Rui XU, Ziwei LIU, Chen Change LOY, Dahua LIN
-
Patent number: 10812825Abstract: The present disclosure provides systems and methods that leverage machine-learned models (e.g., neural networks) to provide video frame synthesis. In particular, the systems and methods of the present disclosure can include or otherwise leverage a machine-learned video frame synthesis model to allow for video frames to be synthesized from videos. In one particular example, the video frame synthesis model can include a convolutional neural network having a voxel flow layer and provides one or more synthesized video frames as part of slow-motion video.Type: GrantFiled: September 29, 2017Date of Patent: October 20, 2020Assignee: Google LLCInventors: Ziwei Liu, Yiming Liu, Aseem Agarwala
-
Patent number: 10761284Abstract: A polyurethane composition is provided. The polyurethane composition includes a first part of a first polytetramethylene oxide, a second polytetramethylene oxide, and a castor oil based polyol. The second polytetramethylene oxide has a higher viscosity than the first polytetramethylene oxide. The polyurethane composition also includes a second part of methylene diphenyl diisocyanate. Also provided is a fiber optic cable assembly incorporating the polyurethane composition as an overmold. The overmold has a glass transition temperature of less than ?40° C. measured according to differential scanning calorimetry. Further provided is a method of applying an overmold to a fiber optic cable assembly.Type: GrantFiled: November 19, 2018Date of Patent: September 1, 2020Assignee: CORNING OPTICAL COMMUNICATIONS LLCInventors: Anil Bastola, Wei Liu, Ziwei Liu
-
Patent number: 10699170Abstract: Disclosed is a method for generating a semantic image labeling model, comprising: forming a first CNN and a second CNN, respectively; randomly initializing the first CNN; inputting a raw image and predetermined label ground truth annotations to the first CNN to iteratively update weights thereof so that a category label probability for the image, which is output from the first CNN, approaches the predetermined label ground truth annotations; randomly initializing the second CNN; inputting the category label probability to the second CNN to correct the input category label probability so as to determine classification errors of the category label probabilities; updating the second CNN by back-propagating the classification errors; concatenating the updated first and second CNNs; classifying each pixel in the raw image into one of general object categories; and back-propagating classification errors through the concatenated CNN to update weights thereof until the classification errors less than a predeterminedType: GrantFiled: January 8, 2018Date of Patent: June 30, 2020Assignee: Beijing SenseTime Technology Development Co., Ltd.Inventors: Xiaoou Tang, Ziwei Liu, Xiaoxiao Li, Ping Luo, Chen Change Loy
-
Publication number: 20200134375Abstract: A semantic segmentation model training method includes: performing, by a semantic segmentation model, image semantic segmentation on at least one unlabeled image to obtain a preliminary semantic segmentation result as the category of the unlabeled image; obtaining, by a convolutional neural network based on the category of the at least one unlabeled image and the category of at least one labeled image, sub-images respectively corresponding to the at least two images and features corresponding to the sub-images, where the at least two images comprise the at least one unlabeled image and the at least one labeled image, and the at least two sub-images carry the categories of the corresponding images; and training the semantic segmentation model on the basis of the categories of the at least two sub-images and feature distances between the at least two sub-images.Type: ApplicationFiled: December 25, 2019Publication date: April 30, 2020Applicant: BEIJING SENSETIME TECHNOLOGY DEVELOPMENT CO., LTD.Inventors: Xiaohang ZHAN, Ziwei LIU, Ping LUO, Chen Change LOY, Xiaoou TANG
-
Publication number: 20190289321Abstract: The present disclosure provides systems and methods that leverage machine-learned models (e.g., neural networks) to provide video frame synthesis. In particular, the systems and methods of the present disclosure can include or otherwise leverage a machine-learned video frame synthesis model to allow for video frames to be synthesized from videos. In one particular example, the video frame synthesis model can include a convolutional neural network having a voxel flow layer and provides one or more synthesized video frames as part of slow-motion video.Type: ApplicationFiled: September 29, 2017Publication date: September 19, 2019Inventors: Ziwei Liu, Yiming Liu, Aseem Agarwala
-
Publication number: 20190138816Abstract: A method and an apparatus for segmenting a video object, an electronic device, a storage medium, and a program include: performing, among at least some frames of a video, inter-frame transfer of an object segmentation result of a reference frame in sequence from the reference frame, to obtain an object segmentation result of at least one other frame among the at least some frames; determining other frames having lost objects with respect to the object segmentation result of the reference frame among the at least some frames; using the determined other frames as target frames to segment the lost objects, so as to update the object segmentation results of the target frames; and transferring the updated object segmentation results of the target frames to the at least one other frame in the video in sequence. The accuracy of video object segmentation results can therefore be improved.Type: ApplicationFiled: December 29, 2018Publication date: May 9, 2019Applicant: BEIJING SENSETIME TECHNOLOGY DEVELOPMENT CO., LTDInventors: Xiaoxiao LI, Yuankai Qi, Zhe Wang, Kai Chen, Ziwei Liu, Jianping Shi, Ping Luo, Chen Change Loy, Xiaoou Tang
-
Publication number: 20190086625Abstract: A polyurethane composition is provided. The polyurethane composition includes a first part of a first polytetramethylene oxide, a second polytetramethylene oxide, and a castor oil based polyol. The second polytetramethylene oxide has a higher viscosity than the first polytetramethylene oxide. The polyurethane composition also includes a second part of methylene diphenyl diisocyanate. Also provided is a fiber optic cable assembly incorporating the polyurethane composition as an overmold. The overmold has a glass transition temperature of less than ?40° C. measured according to differential scanning calorimetry. Further provided is a method of applying an overmold to a fiber optic cable assembly.Type: ApplicationFiled: November 19, 2018Publication date: March 21, 2019Inventors: Anil Bastola, Wei Liu, Ziwei Liu
-
Publication number: 20180129919Abstract: Disclosed is a method for generating a semantic image labeling model, comprising: forming a first CNN and a second CNN, respectively; randomly initializing the first CNN; inputting a raw image and predetermined label ground truth annotations to the first CNN to iteratively update weights thereof so that a category label probability for the image, which is output from the first CNN, approaches the predetermined label ground truth annotations; randomly initializing the second CNN; inputting the category label probability to the second CNN to correct the input category label probability so as to determine classification errors of the category label probabilities; updating the second CNN by back-propagating the classification errors; concatenating the updated first and second CNNs; classifying each pixel in the raw image into one of general object categories; and back-propagating classification errors through the concatenated CNN to update weights thereof until the classification errors less than a predeterminedType: ApplicationFiled: January 8, 2018Publication date: May 10, 2018Inventors: Xiaoou Tang, Ziwei Liu, Xiaoxiao Li, Ping Luo, Chen Change Loy
-
Publication number: 20180045905Abstract: A sealed multiport splitter device and method are disclosed. The fiber optic multiport comprises housing with an interior defining a first chamber and an adjacent second chamber. An optical splitter with a plurality of splitter legs connect to a plurality of optical fibers located in the first chamber. Potting material is disposed in the first chamber to physically secure the optical splitter, the splitter legs and the optical fibers. A wall having a first face and a second face separates the first chamber from the second chamber. The wall has a plurality of slots extended from the first face to the second face. The plurality of optical fibers route through the plurality of slots between the first chamber and the second chamber. A blocking material is adjacent to the wall to inhibit ingress of the potting material into the second chamber while the potting material is being cured.Type: ApplicationFiled: August 12, 2016Publication date: February 15, 2018Inventors: Monique Lise Cote, Troy Dean Lindblad, Ziwei Liu
-
Publication number: 20170315306Abstract: A method of securing an optical fiber to a ferrule includes disposing an adhesive composition in a fiber-receiving passage of the ferrule to form a ferrule adhesion system. The adhesive composition in the ferrule adhesion system comprises a polymeric material in a solid form. The method also involves assembling an optical connector with the ferrule adhesion system, heating the ferrule in the optical connector to melt the adhesive composition, and inserting the optical fiber into the fiber-receiving passage of the ferrule and into contact with the adhesive composition that has been melted. The adhesive composition bonds the optical fiber to an inner surface of the ferrule upon cooling and solidifying.Type: ApplicationFiled: July 17, 2017Publication date: November 2, 2017Inventors: Edward John Fewkes, John Paul Krug, Ziwei Liu
-
Patent number: 9733435Abstract: A method of securing an optical fiber to a ferrule includes: heating an adhesive composition that is disposed within the ferrule to melt the adhesive composition; inserting the optical fiber into a fiber-receiving passage defining an inner surface of the ferrule and into contact with the melted adhesive composition; and cooling the melted adhesive composition. The adhesive composition is a solid material disposed within the fiber-receiving passage of the ferrule and in contact with the inner surface of the ferrule prior to the heating step. Additionally, the adhesive composition comprises a partially cross-linked resin and a coupling agent that covalently bonds the partially cross-linked resin to respective inorganic surfaces of the optical fiber and the ferrule following the cooling step.Type: GrantFiled: November 22, 2016Date of Patent: August 15, 2017Assignee: Corning Optical Communications LLCInventors: Edward John Fewkes, John Paul Krug, Ziwei Liu
-
Publication number: 20170075076Abstract: A method of securing an optical fiber to a ferrule includes: heating an adhesive composition that is disposed within the ferrule to melt the adhesive composition; inserting the optical fiber into a fiber-receiving passage defining an inner surface of the ferrule and into contact with the melted adhesive composition; and cooling the melted adhesive composition. The adhesive composition is a solid material disposed within the fiber-receiving passage of the ferrule and in contact with the inner surface of the ferrule prior to the heating step. Additionally, the adhesive composition comprises a partially cross-linked resin and a coupling agent that covalently bonds the partially cross-linked resin to respective inorganic surfaces of the optical fiber and the ferrule following the cooling step.Type: ApplicationFiled: November 22, 2016Publication date: March 16, 2017Inventors: Edward John Fewkes, John Paul Krug, Ziwei Liu
-
Patent number: 9568686Abstract: An optical connector for terminating an optical fiber includes a ferrule, an optical fiber, and an adhesive composition disposed within a fiber-receiving passage of the ferrule. The adhesive composition, which is in contact with the inner surface of the ferrule and the optical fiber, includes a partially cross-linked resin that is a polymer and a coupling agent that chemically bonds the partially cross-linked resin to an inorganic surface of at least one of the optical fiber and the ferrule.Type: GrantFiled: January 13, 2014Date of Patent: February 14, 2017Assignee: Corning Optical Communications LLCInventors: Edward John Fewkes, John Paul Krug, Ziwei Liu