中科微点-全球专利检索

  1. Login
  2. Sign Up

Search Results

90 records (took 0.015 seconds)

    Video capture of multi-faceted documents
    13.
    发明授权
    Video capture of multi-faceted documents 有权
    视频捕获多面文件

    公开(公告)号：US09241102B2

    公开(公告)日：2016-01-19

    申请号：US13705771

    申请日：2012-12-05

    Applicant: Xerox Corporation

    Inventor： Raja Bala ,  Hengzhou Ding ,  Jayant Kumar

    IPC: H04N5/225 ,  H04N5/232 ,  G06K9/03 ,  G06T3/40 ,  H04N1/387

    CPC classification number: H04N5/23229 ,  G06K9/036 ,  G06K2209/01 ,  G06T3/4092 ,  H04N1/3872 ,  H04N5/23254 ,  H04N5/23267

    Abstract: A system captures or otherwise receives a video and uses the video to create an electronic file corresponding to a multi-faceted printed artifact, such as a multi-page document. When the system receives the video, it selects a set of some or all of the video's image frames, determines a frame quality for each frame in the set, and identifies a subset of the frames such that the frame quality of each frame in the subset satisfies one or more image quality criteria. The subset will include at least one frame for each facet of the multi-faceted printed artifact, such as a page of the document. The processor then automatically combines the subset of frames into a single electronic file.

    Abstract translation: 系统捕获或以其他方式接收视频，并使用该视频来创建对应于多面印刷工件的电子文件，例如多页文档。 当系统接收到视频时，它选择一组视频的图像帧中的一些或全部，确定该组中每个帧的帧质量，并且识别该帧的子集，使得该子集中的每个帧的帧质量 满足一个或多个图像质量标准。 该子集将包括多面印刷工件的每个面的至少一个框架，例如文档的页面。 然后，处理器将帧的子集自动组合成单个电子文件。

    Traffic camera calibration update utilizing scene analysis
    14.
    发明授权
    Traffic camera calibration update utilizing scene analysis 有权
    使用场景分析的交通照相机校准更新

    公开(公告)号：US09185402B2

    公开(公告)日：2015-11-10

    申请号：US13868267

    申请日：2013-04-23

    Applicant: Xerox Corporation

    Inventor： Wencheng Wu ,  Edul N. Dalal ,  Raja Bala

    IPC: H04N17/00 ,  H04N17/02 ,  G06T7/00

    CPC classification number: H04N17/002 ,  G06T7/80 ,  G06T2207/10016 ,  G06T2207/20076 ,  G06T2207/30232 ,  G06T2207/30236

    Abstract: Methods and systems are disclosed for updating camera geometric calibration utilizing scene analysis. Geometric calibration parameters can be derived with respect to one or more cameras and selected reference points of interest identified from a scene acquired by one or more of such cameras. The camera geometric calibration parameters can be applied to image coordinates of the selected reference points of interest to provide real-world coordinates at a time of initial calibration of the camera(s). A subset of a video stream from the camera(s) can then be analyzed to identify features of a current scene captured by the camera(s) that match the selected reference points of interest and provide a current update of the camera geometric calibration parameters with respect to the current scene.

    Abstract translation: 公开了利用场景分析更新摄像机几何校准的方法和系统。 可以相对于从一个或多个这样的相机获取的场景中识别的一个或多个摄像机和所选择的感兴趣的参考点导出几何校准参数。 相机几何校准参数可以应用于所选择的参考点的图像坐标，以在相机的初始校准时提供真实世界坐标。 然后可以分析来自相机的视频流的子集，以识别由相机捕获的当前场景的特征，该特征匹配所选择的感兴趣的参考点，并且将相机几何校准参数的当前更新提供给 尊重当前场景。

    METHODS AND SYSTEMS FOR EFFICIENT IMAGE CROPPING AND ANALYSIS
    15.
    发明申请
    METHODS AND SYSTEMS FOR EFFICIENT IMAGE CROPPING AND ANALYSIS 有权
    高效图像拼接与分析的方法与系统

    公开(公告)号：US20150294175A1

    公开(公告)日：2015-10-15

    申请号：US14249809

    申请日：2014-04-10

    Applicant: Xerox Corporation

    Inventor： Raja Bala ,  Aaron M. Burry

    IPC: G06K9/32 ,  G06T3/40 ,  G06K9/18

    CPC classification number: G06K9/325 ,  G06T3/00 ,  G06T7/73 ,  G06T2207/20021 ,  G06T2207/20132

    Abstract: A system and method for cropping a license plate image to facilitate license plate recognition by obtaining an image that includes the license plate image, dividing the image into multiple sub-blocks, computing an activity measure for each sub-block; determining an activity threshold, determining that a sub-block is an active sub-block by comparing the activity measure for the sub-block with the activity threshold, generating a second image of the license plate information, where the second image includes the active sub-block, and obtaining the license plate information based on the second image.

    Abstract translation: 一种用于通过获取包括车牌图像的图像来划分车牌图像以便于车牌识别的系统和方法，将图像划分成多个子块，计算每个子块的活动度量; 确定活动阈值，通过将子块的活动度量与活动阈值进行比较来确定子块是活动子块，生成车牌信息的第二图像，其中第二图像包括活动子图 块，并且基于第二图像获得车牌信息。

    METHOD AND SYSTEM FOR ESTIMATING GAZE DIRECTION OF VEHICLE DRIVERS
    16.
    发明申请
    METHOD AND SYSTEM FOR ESTIMATING GAZE DIRECTION OF VEHICLE DRIVERS 有权
    估计车辆行驶方向和方法的方法和系统

    公开(公告)号：US20150116493A1

    公开(公告)日：2015-04-30

    申请号：US14274315

    申请日：2014-05-09

    Applicant: Xerox Corporation

    Inventor： Raja Bala ,  Edgar Bernal ,  Aaron Burry ,  Peter Paul ,  David Chuang

    IPC: G06K9/00 ,  B60R1/00

    CPC classification number: G06K9/00845 ,  G06K9/00281 ,  G06K9/6277

    Abstract: Methods and systems for continuously monitoring the gaze direction of a driver of a vehicle over time. Video is received, which is captured by a camera associated with, for example, a mobile device within a vehicle, the camera and/or mobile device mounted facing the driver of the vehicle. Frames can then be extracted from the video. A facial region can then be detected, which corresponds to the face of the driver within the extracted frames. Features descriptors can then be computed from the facial region. A gaze classifier derived from the vehicle, the driver, and the camera can then be applied, wherein the gaze classifier receives the feature descriptors as inputs and outputs at least one label corresponding to one or more predefined finite number of gaze classes to identify the gaze direction of the driver of the vehicle.

    Abstract translation: 随着时间的推移，持续监控车辆驾驶员凝视方向的方法和系统。 接收到视频，其由与例如车辆内的移动设备相关联的摄像机捕获，相机和/或移动设备面向车辆驾驶员安装。 然后可以从视频中提取帧。 然后可以检测到面部区域，其对应于所提取的帧内的驾驶员的脸部。 然后可以从面部区域计算特征描述符。 然后可以应用从车辆，驾驶员和照相机得到的注视分类器，其中注视分类器接收特征描述符作为输入，并且输出与一个或多个预定义的有限数量的注视类别对应的至少一个标签，以识别注视 车辆司机的方向。

    VIDEO CAPTURE OF MULTI-FACETED DOCUMENTS
    17.
    发明申请
    VIDEO CAPTURE OF MULTI-FACETED DOCUMENTS 有权
    多媒体文件的视频捕获

    公开(公告)号：US20140152849A1

    公开(公告)日：2014-06-05

    申请号：US13705771

    申请日：2012-12-05

    Applicant: XEROX CORPORATION

    Inventor： Raja Bala ,  Hengzhou Ding ,  Jayant Kumar

    IPC: H04N5/225

    CPC classification number: H04N5/23229 ,  G06K9/036 ,  G06K2209/01 ,  G06T3/4092 ,  H04N1/3872 ,  H04N5/23254 ,  H04N5/23267

    Abstract: A system captures or otherwise receives a video and uses the video to create an electronic file corresponding to a multi-faceted printed artifact, such as a multi-page document. When the system receives the video, it selects a set of some or all of the video's image frames, determines a frame quality for each frame in the set, and identifies a subset of the frames such that the frame quality of each frame in the subset satisfies one or more image quality criteria. The subset will include at least one frame for each facet of the multi-faceted printed artifact, such as a page of the document. The processor then automatically combines the subset of frames into a single electronic file.

    Abstract translation: 系统捕获或以其他方式接收视频，并使用该视频来创建对应于多面印刷工件的电子文件，例如多页文档。 当系统接收到视频时，它选择一组视频的图像帧中的一些或全部，确定该组中每个帧的帧质量，并且识别该帧的子集，使得该子集中的每个帧的帧质量 满足一个或多个图像质量标准。 该子集将包括多面印刷工件的每个面的至少一个框架，例如文档的页面。 然后，处理器将帧的子集自动组合成单个电子文件。

    USING MULTIPLE TRAINED MODELS TO REDUCE DATA LABELING EFFORTS
    18.
    发明公开
    USING MULTIPLE TRAINED MODELS TO REDUCE DATA LABELING EFFORTS 审中-公开

    公开(公告)号：US20240281431A1

    公开(公告)日：2024-08-22

    申请号：US18647425

    申请日：2024-04-26

    Applicant: Xerox Corporation

    Inventor： Matthew Shreve ,  Francisco E. Torres ,  Raja Bala ,  Robert R. Price ,  Pei Li

    IPC: G06F16/23 ,  G06N20/00

    CPC classification number: G06F16/2379 ,  G06N20/00

    Abstract: A method of labeling training data includes inputting a plurality of unlabeled input data samples into each of a plurality of pre-trained neural networks and extracting a set of feature embeddings from multiple layer depths of each of the plurality of pre-trained neural networks. The method also includes generating a plurality of clusterings from the set of feature embeddings. The method also includes analyzing, by a processing device, the plurality of clusterings to identify a subset of the plurality of unlabeled input data samples that belong to a same unknown class. The method also includes assigning pseudo-labels to the subset of the plurality of unlabeled input data samples.

Patent Agency Ranking