专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

1. 发明公开

EP3341891A1 AKTIVES MASCHINELLES LERNEN ZUM TRAINIEREN EINES EREIGNISKLASSIFIKATORS 无效
公开(公告)号：EP3341891A1
公开(公告)日：2018-07-04
申请号：EP16763215.7
申请日：2016-08-23
申请人： Carl Zeiss AG , Carl Zeiss Microscopy GmbH
发明人： KANDEMIR, Melih , HAMPRECHT, Fred , WOJEK, Christian , SCHMIDT, Ute
IPC分类号： G06K9/62
CPC分类号： G06K9/6263 , G06K9/6256 , G06K9/6278 , G06K9/6284
摘要： An event classification (131) is trained by means of machine learning. To this end, an anomaly detection (121) for detecting events in an image data set (110, 111) is carried out. Based on the performing of the anomaly detection (121), a model assumption (130) of the event classification (131) is determined.

2. 发明公开

EP3332356A1 SEMI-AUTOMATIC IMAGE SEGMENTATION 有权转让
公开(公告)号：EP3332356A1
公开(公告)日：2018-06-13
申请号：EP16828308.3
申请日：2016-07-15
申请人： Sony Corporation
发明人： TAKEDA, Hiroyuki , GHARAVI-ALKHANSARI, Mohammad
IPC分类号： G06K9/34
CPC分类号： G06F3/04842 , G06K9/00234 , G06K9/6253 , G06K9/6278 , G06K9/6857 , G06T7/11 , G06T2207/10024 , G06T2207/20016 , G06T2207/20076 , G06T2207/20101 , G06T2207/30201 , G06T2207/30232
摘要： Semi-automatic (instant) color image segmentation is performed on an input image for an object located near the point given by the user (e.g., touch screen contact). The size, shape, and the number of colors of the target object which is about (lying around or surrounding) the user-selected point is determined, and determinations made for which region the pixels belong to, either foreground or background. A binary object mask is generated which can be utilized for removing the material surrounding that object, or inverting the mask to remove the material of the object.

3. 发明公开

EP3304436A1 FAST LOW-MEMORY METHODS FOR BAYESIAN INFERENCE, GIBBS SAMPLING AND DEEP LEARNING 审中-公开
标题翻译：用于贝叶斯推理，GIBBS采样和深度学习的快速低内存方法
公开(公告)号：EP3304436A1
公开(公告)日：2018-04-11
申请号：EP16728149.2
申请日：2016-05-18
申请人： Microsoft Technology Licensing, LLC
发明人： WIEBE, Nathan , KAPOOR, Ashish , SVORE, Krysta , GRANADE, Christopher
IPC分类号： G06N3/04 , G06N7/00 , G06F17/10 , G06N99/00
CPC分类号： G06N5/022 , G06K9/6226 , G06K9/6256 , G06K9/6278 , G06N3/0445 , G06N7/005 , G06N99/002 , G06N99/005
摘要： Methods of training Boltzmann machines include rejection sampling to approximate a Gibbs distribution associated with layers of the Boltzmann machine. Accepted sample values obtained using a set of training vectors and a set of model values associate with a model distribution are processed to obtain gradients of an objective function so that the Boltzmann machine specification can be updated. In other examples, a Gibbs distribution is estimated or a quantum circuit is specified so at to produce eigenphases of a unitary.

4. 发明公开

EP3229171A1 METHOD AND DEVICE FOR DETERMINING IDENTITY IDENTIFIER OF HUMAN FACE IN HUMAN FACE IMAGE, AND TERMINAL 审中-公开
标题翻译：用于确定人脸图像中人脸的身份标识的方法和装置以及终端
公开(公告)号：EP3229171A1
公开(公告)日：2017-10-11
申请号：EP15875160.2
申请日：2015-12-24
申请人： Huawei Technologies Co. Ltd.
发明人： JU, Wenqi , LI, Wei , XU, Chunjing
IPC分类号： G06K9/00
CPC分类号： G06K9/00288 , G06K9/00 , G06K9/00268 , G06K9/6278
摘要： The present invention provides a method and an apparatus for determining an identity identifier of a face in a face image, and a terminal. The method includes: obtaining an original feature vector of a face image; selecting k candidate vectors from a face image database according to the original feature vector; selecting a matching vector of the original feature vector from the k candidate vectors; and determining, according to the matching vector of the original feature vector, an identity identifier that is of the matching vector and that is recorded in the face image database as an identity identifier of a face in the face image. In embodiments of the present invention, a face image database stores a medium-level feature vector formed by means of mutual interaction between a low-level face feature vector and autocorrelation and cross-correlation submatrices in a joint Bayesian probability matrix. The medium-level feature vector includes information about mutual interaction between the face feature vector and the autocorrelation and cross-correlation submatrices in the joint Bayesian probability matrix, so that efficiency and accuracy of facial recognition can be improved.
摘要翻译：本发明提供了一种确定人脸图像中人脸身份标识的方法及装置，以及终端。该方法包括：获取人脸图像的原始特征向量; 根据原始特征向量从人脸图像数据库中选择k个候选向量; 从k个候选矢量中选择原始特征矢量的匹配矢量; 根据所述原始特征向量的匹配向量确定所述匹配向量的身份标识，所述身份标识记录在所述人脸图像数据库中，作为所述人脸图像中人脸的身份标识。在本发明的实施例中，人脸图像数据库存储通过联合贝叶斯概率矩阵中的低级脸部特征向量与自相关和互相关子矩阵之间的相互作用形成的中等特征向量。中等特征矢量包括关于人脸特征向量与联合贝叶斯概率矩阵中的自相关和互相关子矩阵之间的相互作用的信息，从而可以提高面部识别的效率和准确性。

5. 发明授权

EP2977956B1 METHOD, APPARATUS AND DEVICE FOR SEGMENTING AN IMAGE 有权
标题翻译： VERFAHREN，VORRICHTUNG UND EINRICHTUNG ZUR SEGMENTIERUNG EINES BILDES
公开(公告)号：EP2977956B1
公开(公告)日：2017-01-11
申请号：EP15175108.8
申请日：2015-07-02
申请人： Xiaomi Inc.
发明人： WANG, Lin , XU, Xiaozhou , CHEN, Zhijun
IPC分类号： G06T7/00 , G06K9/00
CPC分类号： G06T7/143 , G06K9/00234 , G06K9/00268 , G06K9/00288 , G06K9/4604 , G06K9/4652 , G06K9/52 , G06K9/6218 , G06K9/6278 , G06T7/11 , G06T7/12 , G06T7/136 , G06T7/194 , G06T7/73 , G06T2207/30201
摘要： The present disclosure relates to a method, apparatus and device for segmenting image. The method includes: calculating an a priori probability of appearing a head-shoulder foreground at each pixel in an image having a preset size; selecting head-shoulder foreground and background sample pixels from an image to be segmented according to the a priori probability, and a head-shoulder foreground and a background probability threshold set in advance; calculating a first color likelihood probability of the head-shoulder foreground and a second color likelihood probability of the background according to color feature vectors of the head-shoulder foreground and the background sample pixels; calculating a first a posteriori probability of the head-shoulder foreground and a second a posteriori probability of the background according to the a priori probability, first and second color likelihood probabilities; and performing head-shoulder segmentation on the image according to the first and second a posteriori probabilities. The segmentation precision is high.
摘要翻译：本公开涉及用于分割图像的方法，装置和装置。该方法包括：计算在具有预设大小的图像中的每个像素处出现头肩前景的先验概率; 从根据先验概率分割的图像中选择头肩前景和背景样本像素，以及预先设置的头肩前景和背景概率阈值; 根据头肩前景和背景样本像素的颜色特征向量计算头肩前景的第一颜色似然概率和背景的第二颜色似然概率; 根据先验概率，第一和第二颜色似然概率计算头肩前景的第一后验概率和背景的第二后验概率; 以及根据第一和第二后验概率在图像上执行头肩分割。分割精度高。

6. 发明公开

EP2340525A4 DETECTION OF VEHICLES IN AN IMAGE 审中-公开
标题翻译：在EINEM BILD中检测VON FAHRZEUGEN
公开(公告)号：EP2340525A4
公开(公告)日：2016-12-07
申请号：EP09815486
申请日：2009-09-25
申请人： NAT ICT AUSTRALIA LTD
发明人： ROBERT KOSTIA , YANG JUN , WANG YANG , ZHANG JIAN
IPC分类号： G06T7/60 , G06K9/62
CPC分类号： G06K9/6278 , G06K9/00785 , G06K9/3241 , G06K9/6204 , G06K2009/6213 , G06K2209/23 , G06T7/12 , G06T7/149 , G06T2207/30236

7. 发明公开

EP3082051A1 DATA MINING METHOD 审中-公开
标题翻译： DATENAUSWERTUNGSVERFAHREN
公开(公告)号：EP3082051A1
公开(公告)日：2016-10-19
申请号：EP14869820.2
申请日：2014-12-10
申请人： China Unionpay Co., Ltd.
发明人： WANG, Jun , YANG, Hongchao
IPC分类号： G06F17/30
CPC分类号： G06F17/30539 , G06F17/30536 , G06F17/30598 , G06K9/6267 , G06K9/6278 , G06K9/6284
摘要： The present invention proposes a method for data mining, the method comprising: making statistics of the feature vectors of each target object according to the records in a target data set so as to constitute a rough data set, each of the feature vectors including the value of at least one attribute data of the target objects corresponding thereto; screening the feature vectors which correspond to all known the first type of target objects from the rough data set, and performing a filter operation onto the screened feature vectors to obtain samples; and building a regression model based on the samples, and then using the built regression model to determine whether each of all known the second type of target objects potentially belongs to the first type of target objects. The method for data mining disclosed in the present invention is capable of mining and classifying the target objects according to the comprehensive features of the target objects.
摘要翻译：本发明提出了一种数据挖掘方法，该方法包括：根据目标数据集中的记录，对每个目标对象的特征向量进行统计，构成粗略数据集，每个特征向量包括值与其对应的目标对象的至少一个属性数据; 从粗略数据集筛选与所有已知的第一类型的目标对象相对应的特征向量，并对筛选的特征向量执行过滤操作以获得样本; 并基于样本构建回归模型，然后使用内建的回归模型来确定所有已知的第二类型的目标对象中的每一个是否潜在地属于第一类型的目标对象。本发明公开的数据挖掘方法能够根据目标对象的综合特征挖掘和分类目标对象。

8. 发明公开

EP2983111A3 METHOD AND APPARATUS FOR FACIAL RECOGNITION 审中-实审
标题翻译： GESICHTSKENNUNGSVERFAHREN UND -VORRICHTUNG
公开(公告)号：EP2983111A3
公开(公告)日：2016-04-13
申请号：EP15179878.2
申请日：2015-08-05
申请人： Samsung Electronics Co., Ltd
发明人： SUH, SUNGJOO , HWANG, Wonjun , KIM, Wonjun , HAN, JAEJOON
IPC分类号： G06K9/00 , G06K9/62
CPC分类号： G06K9/00221 , G06K9/00228 , G06K9/00248 , G06K9/00281 , G06K9/00288 , G06K9/6209 , G06K9/6278 , G06K9/6296 , G06K2209/21
摘要： At least some example embodiments disclose a method and apparatus for facial recognition. The facial recognition method includes detecting initial landmarks from a facial image, first normalizing the facial image using the initial landmarks, updating a position of at least one of intermediate landmarks based on the first normalizing, the intermediate landmarks being landmarks transformed from the initial landmarks through the first normalizing, second normalizing the facial image after the updating and recognizing a face using a feature of the second normalized facial image.
摘要翻译：至少一些示例性实施例公开了用于面部识别的方法和装置。面部识别方法包括从面部图像检测初始界标，首先使用初始地标标准化面部图像，基于第一归一化来更新中间地标中的至少一个的位置，中间地标是从初始地标变换的地标，通过第一归一化，使用第二标准化脸部图像的特征更新并识别脸部之后对面部图像进行二次归一化。

9. 发明公开

EP2581848A3 Clinical information processing apparatus, method and program 审中-公开
标题翻译：一种装置，方法和程序用于处理临床信息
公开(公告)号：EP2581848A3
公开(公告)日：2014-07-30
申请号：EP12188315.1
申请日：2012-10-12
申请人： Fujifilm Corporation
发明人： Kanada, Shoji
IPC分类号： G06F19/00 , G06K9/62
CPC分类号： G06F19/3443 , G06F19/00 , G06F19/321 , G06K9/6278 , G16H50/70
摘要： Even if the number of comparison target patients' cases is small, weighting is appropriately performed based on clinical-information items, and a degree of similarity between a target patient's case and a case of a comparison target patient is calculated. Likelihood ratio between a likelihood of belonging to one classification of a key item and a likelihood of belonging to other classification of the key item when a case belongs to each classification of a clinical-information item other than the key item is calculated, based on registration case information for calculating a likelihood ratio, for each classification of a key item. A weighting coefficient corresponding to each classification of the clinical-information item other than the key item for each classification of the key item is determined based on a target classification of a target clinical-information item and the calculated likelihood ratio. A degree of similarity is calculated for each registration case included in registration case information for calculating a degree of similarity by using weighting information corresponding to each classification of the key item and each classification of the clinical-information item other than the key item.

10. 发明公开

EP2348458A1 Biometric authentication system 有权
标题翻译： Biometrisches认证系统
公开(公告)号：EP2348458A1
公开(公告)日：2011-07-27
申请号：EP10251554.1
申请日：2010-09-03
申请人： Hitachi, Ltd.
发明人： Takao Murakam , Kenta Takahashi
IPC分类号： G06K9/68 , G06K9/62
CPC分类号： G06K9/6278 , G06K9/00885 , G06K9/68
摘要： A 1:N identification system having high convenience and safety is to be provided. An authentication client (100) includes at least one biometric information input sensor (101) and a feature extraction function (102). A database (120) includes an enrollee ID (122) and registered templates (123) of biometric information of at least one kind every enrollee and includes a score table (124). An authentication server (110) includes a prior probability setting function (111), a 1:N fast matching function (112) for successively matching the feature with the registered templates of the enrollees and discontinuing matching processing when the number of times of matching has exceeded a predetermined threshold, a delta score calculation function (113) for calculating a delta score by using a score obtained by the 1:N fast matching and using the score table, a posterior probability calculation function (114) for calculating posterior probabilities respectively of the enrollees on the basis of the score and the delta score, and an authentication object user identification function (115).
摘要翻译：提供一种具有高便利性和安全性的1：N识别系统。认证客户端（100）包括至少一个生物特征信息输入传感器（101）和特征提取功能（102）。数据库（120）包括每个登记者至少一种类型的生物体信息的登记者ID（122）和注册模板（123），并且包括分数表（124）。认证服务器（110）包括先验概率设置功能（111），1：N快速匹配功能（112），用于在特征与注册者的注册模板连续匹配并且当匹配次数为超过预定阈值的增量分数计算功能（113），用于通过使用通过1：N快速匹配获得的分数并使用分数表来计算增量分数;后验概率计算功能（114），用于分别计算后验概率基于得分和增量得分的参与者，以及认证对象用户识别功能（115）。

你已经成功收藏专利！

检索式保存成功!

IPRDB

热门服务

关于我们

友情链接

联系方式