-
公开(公告)号：US09767349B1
公开(公告)日：2017-09-19
申请号：US15149284
申请日：2016-05-09
Applicant: Xerox Corporation
Inventor： Matthew Adam Shreve , Jayant Kumar , Raja Bala , Phillip J. Emmett , Megan Clar , Jeyasri Subramanian , Eric Harte
CPC classification number: G06K9/00335 , G06K9/00302 , G06K9/00315
Abstract: A method for determining an emotional state of a subject taking an assessment. The method includes eliciting predicted facial expressions from a subject administered questions each intended to elicit a certain facial expression that conveys a baseline characteristic of the subject; receiving a video sequence capturing the subject answering the questions; determining an observable physical behavior experienced by the subject across a series of frames corresponding to the sample question; associating the observed behavior with the emotional state that corresponds with the facial expression; and training a classifier using the associations. The method includes receiving a second video sequence capturing the subject during an assessment and applying features extracted from the second image data to the classifier for determining the emotional state of the subject in response to an assessment item administered during the assessment.
-
公开(公告)号：US09392185B1
公开(公告)日：2016-07-12
申请号：US14619956
申请日：2015-02-11
Applicant: Xerox Corporation
Inventor： Jayant Kumar , Sungmin Eum , Raja Bala
CPC classification number: H04N5/2354 , G06T3/4038 , G06T5/005 , G06T5/50 , H04N1/3876 , H04N5/2256 , H04N5/23229 , H04N5/265
Abstract: An apparatus, method and non-transitory computer readable medium for mosaicking a plurality of images captured in a low-light condition are disclosed. For example, the apparatus includes an image capturing module, a flash module, a computer-readable medium and a processor. The processor executes a plurality of instructions stored on the computer-readable medium. The operations include determining a spacing between each one of the plurality of images, capturing the plurality of images in accordance with the spacing, aligning the plurality of images that are captured and mosaicking the plurality of images that are aligned into a single image by, for each one of the plurality of images, replacing one or more pixels in the respective flash spot region using one or more pixels in a subsequent image that is in a same location as the respective flash spot region such that the information is recovered.
Abstract translation: 公开了一种用于镶嵌在低光条件下拍摄的多个图像的装置，方法和非暂时计算机可读介质。 例如，该装置包括图像捕获模块，闪存模块，计算机可读介质和处理器。 处理器执行存储在计算机可读介质上的多个指令。 操作包括确定多个图像中的每个图像之间的间隔，根据间隔捕获多个图像，将捕获的多个图像对齐并将对准的单个图像的多个图像镶嵌在一起，用于 多个图像中的每一个，使用与相应闪光斑区域位于相同位置的后续图像中的一个或多个像素替换相应闪光斑区域中的一个或多个像素，使得信息被恢复。
-
公开(公告)号：US10408912B2
公开(公告)日：2019-09-10
申请号：US15933601
申请日：2018-03-23
Applicant: Xerox Corporation
Inventor： Jayant Kumar , Qun Li , Edgar A. Bernal , Raja Bala
Abstract: A mobile electronic device processes a sequence of images to identify and re-identify an object of interest in the sequence. An image sensor of the device, receives a sequence of images. The device detects an object in a first image as well as positional parameters of the device that correspond to the object in the first image. The device determines a range of positional parameters within which the object may appear in a field of view of the device. When the device detects that the object of interest exited the field of view it subsequently uses motion sensor data to determine that the object of interest has likely re-entered the field of view, it will analyze the current frame to confirm that the object of interest has re-entered the field of view.
-
4.发明授权公开(公告)号：US09807269B2
公开(公告)日：2017-10-31
申请号：US14713077
申请日：2015-05-15
Applicant: Xerox Corporation
Inventor： Jayant Kumar , Raja Bala , Martin S. Maltz , Phillip J Emmett
CPC classification number: H04N1/00997 , G06T5/007 , G06T5/50 , G06T2207/20221 , G06T2207/30176 , H04N1/028
Abstract: The embodiments include systems and methods for guiding a user to capture two flash images of a document page, and selectively fuse the images to produce a binary image of high quality and without loss of any content. Each individual image may have an FSR where the content is degraded/lost due to the flash light. The idea is to first guide the user to take two images such that there is no overlap of flash-spots in the document regions. The flash spots in both images are detected and assessed for quality and extent of degradation in both images. The image with lower degradation is chosen as the primary image and the other image as secondary, to minimize fusing artifacts. The region in secondary image corresponding to the FSR in the primary is aligned to the primary region using a multiscale alignment technique. The primary image and aligned FSR are binarized and fused in the vicinity of the flashspot in the primary using an intelligent technique that minimizes fusion boundary artifacts such as cutting of characters and words.
-
5.发明申请公开(公告)号：US20170255831A1
公开(公告)日：2017-09-07
申请号：US15061463
申请日：2016-03-04
Applicant: Xerox Corporation
Inventor： Edgar A. Bernal , Qun Li , Yun Zhang , Jayant Kumar , Raja Bala
CPC classification number: G06K9/00751 , G06K9/00718 , G06K9/00765 , G06K9/6227 , G06K9/6269 , H04L65/4084 , H04L65/605 , H04L67/04 , H04L67/42 , H04L69/16
Abstract: A method and system for identifying content relevance comprises acquiring video data, mapping the acquired video data to a feature space to obtain a feature representation of the video data, assigning the acquired video data to at least one action class based on the feature representation of the video data, and determining a relevance of the acquired video data.
-
Patent Agency Ranking
公开(公告)号：US20170140253A1
公开(公告)日：2017-05-18
申请号：US15179403
申请日：2016-06-10
Applicant: Xerox Corporation
Inventor： Safwan Wshah , Beilei Xu , Orhan Bulan , Jayant Kumar , Peter Paul
CPC classification number: G06N3/08 , G06K9/00771 , G06K9/00785 , G06K9/4628 , G06K9/629 , G06K2209/23 , G06N3/0454
Abstract: A method and system for domain adaptation based on multi-layer fusion in a convolutional neural network architecture for feature extraction and a two-step training and fine-tuning scheme. The architecture concatenates features extracted at different depths of the network to form a fully connected layer before the classification step. First, the network is trained with a large set of images from a source domain as a feature extractor. Second, for each new domain (including the source domain), the classification step is fine-tuned with images collected from the corresponding site. The features from different depths are concatenated with and fine-tuned with weights adjusted for a specific task. The architecture is used for classifying high occupancy vehicle images.
-
7.发明申请公开(公告)号：US20170061202A1
公开(公告)日：2017-03-02
申请号：US14840745
申请日：2015-08-31
Applicant: Xerox Corporation
Inventor： Matthew Adam Shreve , Jayant Kumar , Qun Li , Edgar A. Bernal , Raja Bala
CPC classification number: G06K9/00315 , G06K9/00758 , G06K9/623
Abstract: A method, computer readable medium and apparatus for verifying an identity of an individual based upon facial expressions as exhibited in a query video of the individual are disclosed. The method includes receiving a reference video for each one of a plurality of different individuals, wherein a plurality of facial gesture encoders is extracted from at least one frame of the reference video describing one or more facial expressions of each one of the plurality of different individuals, receiving the query video, calculating a similarity score for the reference video for the each one of the plurality of different individuals based on an analysis that compares the plurality of facial gesture encoders of the at least one frame of the reference video for the each one of the plurality of different individuals to a plurality of facial gesture encoders extracted from at least one frame of the query video.
Abstract translation: 公开了一种基于个人的查询视频中显示的基于面部表情来验证个人身份的方法，计算机可读介质和装置。 该方法包括接收多个不同个体中的每一个的参考视频，其中从描述多个不同个体中的每一个的一个或多个面部表情的参考视频的至少一个帧中提取多个面部手势编码器 接收查询视频，基于比较每个参考视频的至少一帧的多个面部手势编码器的分析，计算多个不同个体中的每个个体的参考视频的相似性得分 将所述多个不同个体的多个人脸手势编码器从所述查询视频的至少一个帧中提取出来。
-
8.发明申请公开(公告)号：US20180217223A1
公开(公告)日：2018-08-02
申请号：US15933601
申请日：2018-03-23
Applicant: Xerox Corporation
Inventor： Jayant Kumar , Qun Li , Edgar A. Bernal , Raja Bala
CPC classification number: G01S3/7864 , G06K9/00771 , G06K2009/3291 , G06T7/20
Abstract: A mobile electronic device processes a sequence of images to identify and re-identify an object of interest in the sequence. An image sensor of the device, receives a sequence of images. The device detects an object in a first image as well as positional parameters of the device that correspond to the object in the first image. The device determines a range of positional parameters within which the object may appear in a field of view of the device. When the device detects that the object of interest exited the field of view it subsequently uses motion sensor data to determine that the object of interest has likely re-entered the field of view, it will analyze the current frame to confirm that the object of interest has re-entered the field of view.
-
公开(公告)号：US09864931B2
公开(公告)日：2018-01-09
申请号：US15097791
申请日：2016-04-13
Applicant: Xerox Corporation
Inventor： Jayant Kumar , Beilei Xu , Peter Paul
IPC: G06K9/62
CPC classification number: G06K9/6245 , G06K9/6259 , G06K9/6269
Abstract: Methods, systems, and processor-readable media for training data augmentation. A source domain and a target domain are provided, and thereafter an operation is performed to augment data in the source domain with transformations utilizing characteristics learned from the target domain. The augmented data is then used to improve image classification accuracy in a new domain.
-
公开(公告)号：US20170300783A1
公开(公告)日：2017-10-19
申请号：US15097791
申请日：2016-04-13
Applicant: Xerox Corporation
Inventor： Jayant Kumar , Beilei Xu , Peter Paul
IPC: G06K9/62
CPC classification number: G06K9/6245 , G06K9/6259 , G06K9/6269
Abstract: Methods, systems, and processor-readable media for training data augmentation. A source domain and a target domain are provided, and thereafter an operation is performed to augment data in the source domain with transformations utilizing characteristics learned from the target domain. The augmented data is then used to improve image classification accuracy in a new domain.
-
-
-
-
-
-
-
-
-
Search Results
20
records
(took 0.105 seconds)
