-
公开(公告)号：US10147459B2
公开(公告)日：2018-12-04
申请号：US15273695
申请日：2016-09-22
Applicant: Apple Inc.
Inventor： Bartlomiej W. Rymkowski , Marco Zuliani
IPC: G11B27/00 , H04N5/93 , G11B27/031 , G06T11/60 , G06K9/00
Abstract: Techniques are disclosed herein for applying an artistic style extracted from one or more source images, e.g., paintings, to one or more target images. The extracted artistic style may then be stored as a plurality of layers in a neural network. In some embodiments, two or more stylized target images may be combined and stored as a stylized video sequence. The artistic style may be applied to the target images in the stylized video sequence using various optimization methods and/or pixel- and feature-based regularization techniques in a way that prevents excessive content pixel fluctuations between images and preserves smoothness in the assembled stylized video sequence. In other embodiments, a user may be able to semantically annotate locations of undesired artifacts in a target image, as well as portion(s) of a source image from which a style may be extracted and used to replace the undesired artifacts in the target image.
-
公开(公告)号：US20180088996A1
公开(公告)日：2018-03-29
申请号：US15711781
申请日：2017-09-21
Applicant: Apple Inc.
Inventor： Francesco Rossi , Marco Zuliani
IPC: G06F9/50
CPC classification number: G06F9/5016
Abstract: A method may include accessing a data processing architecture associated with a neural network to determine dependencies between intermediate data layers of the neural network; obtaining dimensions of the intermediate data layers in the neural network; calculating a minimum number of data storage portions for executing the neural network based on the dependencies; determining a memory allocation size for each respective data storage portion of the data storage portions based on the dimensions and dependencies; allocating memory on a storage device for each data storage portion in accordance with its respective determined memory allocation size.
-
公开(公告)号：US20180082715A1
公开(公告)日：2018-03-22
申请号：US15273695
申请日：2016-09-22
Applicant: Apple Inc.
Inventor： Bartlomiej W. Rymkowski , Marco Zuliani
IPC: G11B27/031 , G06T11/60 , G06K9/00
CPC classification number: G11B27/031 , G06K9/00664 , G06K9/00718 , G06T11/60
Abstract: Techniques are disclosed herein for applying an artistic style extracted from one or more source images, e.g., paintings, to one or more target images. The extracted artistic style may then be stored as a plurality of layers in a neural network. In some embodiments, two or more stylized target images may be combined and stored as a stylized video sequence. The artistic style may be applied to the target images in the stylized video sequence using various optimization methods and/or pixel- and feature-based regularization techniques in a way that prevents excessive content pixel fluctuations between images and preserves smoothness in the assembled stylized video sequence. In other embodiments, a user may be able to semantically annotate locations of undesired artifacts in a target image, as well as portion(s) of a source image from which a style may be extracted and used to replace the undesired artifacts in the target image.
-
公开(公告)号：US09773192B2
公开(公告)日：2017-09-26
申请号：US14732738
申请日：2015-06-07
Applicant: Apple Inc.
Inventor： Xiaoxing Li , Geoffrey T. Anneheim , Jianping Zhou , Richard D. Seely , Marco Zuliani
CPC classification number: G06K9/6202 , G06T3/40 , G06T7/13 , G06T7/248 , G06T7/90 , G06T2207/10016 , G06T2207/10024
Abstract: Techniques to identify and track a pre-identified region-of-interest (ROI) through a temporal sequence of frames/images are described. In general, a down-sampled color gradient (edge map) of an arbitrary sized ROI from a prior frame may be used to generate a small template. This initial template may be used to identify a region of a new or current frame that may be overscan and used to create a current frame's edge map. By comparing the prior frame's template to the current frame's edge map, a cost value or image may be found and used to identify the current frame's ROI center. The size of the current frame's ROI may be found by varying the size of putative new ROIs and testing for their congruence with the prior frame's template. Subsequent ROI's for subsequent frames may be identified to, effectively, track an arbitrarily sized ROI through a sequence of video frames.
-
公开(公告)号：US09542585B2
公开(公告)日：2017-01-10
申请号：US13911983
申请日：2013-06-06
Applicant: Apple Inc.
Inventor： George Williams , Benjamin Olson , Sebastien Beysserie , Ethan Tira-Thompson , Jianping Zhou , Claus Molgaard , Todd Sachs , Rudolph van der Merwe , Marco Zuliani
CPC classification number: G06K7/1443 , G06K7/10722 , G06K9/3233 , G06K9/6203 , G06K2009/3291 , G06K2209/21 , G06T7/20
Abstract: A method to improve the efficiency of the detection and tracking of machine-readable objects is disclosed. The properties of image frames may be pre-evaluated to determine whether a machine-readable object, even if present in the image frames, would be likely to be detected. After it is determined that one or more image frames have properties that may enable the detection of a machine-readable object, image data may be evaluated to detect the machine-readable object. When a machine-readable object is detected, the location of the machine-readable object in a subsequent frame may be determined based on a translation metric between the image frame in which the object was identified and the subsequent frame rather than a detection of the object in the subsequent frame. The translation metric may be identified based on an evaluation of image data and/or motion sensor data associated with the image frames.
Abstract translation: 公开了一种提高机器可读对象检测和跟踪效率的方法。 可以预先评估图像帧的属性，以确定即使存在于图像帧中的机器可读对象是否可能被检测到。 在确定一个或多个图像帧具有能够检测机器可读对象的属性之后，可以评估图像数据以检测机器可读对象。 当检测到机器可读对象时，可以基于在其中识别对象的图像帧与后续帧之间的转换度量而不是对象的检测来确定随后帧中的机器可读对象的位置 在后续的框架。 可以基于与图像帧相关联的图像数据和/或运动传感器数据的评估来识别翻译度量。
-
Patent Agency Ranking
26.发明申请公开(公告)号：US20150071547A1
公开(公告)日：2015-03-12
申请号：US14021857
申请日：2013-09-09
Applicant: Apple Inc.
Inventor： Brett Keating , Vincent Wong , Todd Sachs , Claus Molgaard , Michael Rousson , Elliott Harris , Justin Titi , Karl Hsu , Jeff Brasket , Marco Zuliani
CPC classification number: H04N5/77 , G06K9/00765 , G06K9/46 , G06K9/6201 , G06K9/6267 , G06K2009/4666 , G06T7/11 , H04N5/147 , H04N5/23219 , H04N5/23222 , H04N5/23229 , H04N5/23293 , H04N21/8456
Abstract: Systems and methods for improving automatic selection of keeper images from a commonly captured set of images are described. A combination of image type identification and image quality metrics may be used to identify one or more images in the set as keeper images. Image type identification may be used to categorize the captured images into, for example, three or more categories. The categories may include portrait, action, or “other.” Depending on the category identified, the images may be analyzed differently to identify keeper images. For portrait images, an operation may be used to identify the best set of faces. For action images, the set may be divided into sections such that keeper images selected from each section tell the story of the action. For the “other” category, the images may be analyzed such that those having higher quality metrics for an identified region of interest are selected.
Abstract translation: 描述了用于改善从一组共同拍摄的图像中保持图像的自动选择的系统和方法。 可以使用图像类型识别和图像质量度量的组合来识别该集合中的一个或多个图像作为保持器图像。 可以使用图像类型识别来将捕获的图像分类为例如三个或更多个类别。 类别可以包括纵向，动作或“其他”。根据所识别的类别，可以不同地分析图像以识别守护者图像。 对于肖像图像，可以使用操作来识别最佳面部组。 对于动作图像，该集合可以被划分为部分，使得从每个部分中选择的守护者图像讲述动作的故事。 对于“其他”类别，可以分析图像，以便选择对于所识别的感兴趣区域具有较高质量度量的图像。
-
27.发明申请公开(公告)号：US20150035991A1
公开(公告)日：2015-02-05
申请号：US14513046
申请日：2014-10-13
Applicant: Apple Inc.
Inventor： Todd S. Sachs , Kevin L. Hunter , Marco Zuliani , Claus Molgaard
CPC classification number: G06T7/0018 , G06T7/80 , G06T2207/10004 , H04N5/23216 , H04N5/23293 , H04N17/002
Abstract: A method for dynamically calibrating rotational offset in a device includes obtaining an image captured by a camera of the device. Orientation information of the device at the time of image capture may be associated with the image. Pixel data of the image may be analyzed to determine an image orientation angle for the image. A device orientation angle may be determined from the orientation information. A rotational offset, based on the image orientation angle and the device orientation angle, may be determined. The rotational offset is relative to the camera or orientation sensor. A rotational bias may be determined from statistical analysis of numerous rotational offsets from numerous respective images. In some embodiments, various thresholds and predetermined ranges may be used to exclude some rotational offsets from the statistical analysis or to discontinue processing for that image.
Abstract translation: 用于动态校准装置中的旋转偏移的方法包括获得由装置的照相机拍摄的图像。 图像拍摄时设备的方向信息可能与图像相关联。 可以分析图像的像素数据以确定图像的图像取向角度。 可以根据取向信息来确定设备取向角度。 可以确定基于图像取向角度和装置取向角度的旋转偏移。 旋转偏移量相对于相机或方位传感器。 旋转偏差可以根据来自大量各自图像的许多旋转偏移的统计分析来确定。 在一些实施例中，可以使用各种阈值和预定范围来从统计分析中排除一些旋转偏移或者中断对该图像的处理。
-
公开(公告)号：US20140363044A1
公开(公告)日：2014-12-11
申请号：US13911983
申请日：2013-06-06
Applicant: Apple Inc.
Inventor： George Williams , Benjamin Olson , Sebastien Beysserie , Ethan Tira-Thompson , Jianping Zhou , Claus Molgaard , Todd Sachs , Rudolph van der Merwe , Marco Zuliani
IPC: G06K9/00
CPC classification number: G06K7/1443 , G06K7/10722 , G06K9/3233 , G06K9/6203 , G06K2009/3291 , G06K2209/21 , G06T7/20
Abstract: A method to improve the efficiency of the detection and tracking of machine-readable objects is disclosed. The properties of image frames may be pre-evaluated to determine whether a machine-readable object, even if present in the image frames, would be likely to be detected. After it is determined that one or more image frames have properties that may enable the detection of a machine-readable object, image data may be evaluated to detect the machine-readable object. When a machine-readable object is detected, the location of the machine-readable object in a subsequent frame may be determined based on a translation metric between the image frame in which the object was identified and the subsequent frame rather than a detection of the object in the subsequent frame. The translation metric may be identified based on an evaluation of image data and/or motion sensor data associated with the image frames.
Abstract translation: 公开了一种提高机器可读对象检测和跟踪效率的方法。 可以预先评估图像帧的属性，以确定即使存在于图像帧中的机器可读对象是否可能被检测到。 在确定一个或多个图像帧具有能够检测机器可读对象的属性之后，可以评估图像数据以检测机器可读对象。 当检测到机器可读对象时，可以基于在其中识别对象的图像帧与后续帧之间的转换度量而不是对象的检测来确定随后帧中的机器可读对象的位置 在后续的框架。 可以基于与图像帧相关联的图像数据和/或运动传感器数据的评估来识别翻译度量。
-
公开(公告)号：US08860818B1
公开(公告)日：2014-10-14
申请号：US13955467
申请日：2013-07-31
Applicant: Apple Inc.
Inventor： Todd S. Sachs , Kevin L. Hunter , Marco Zuliani , Claus Molgaard
CPC classification number: G06T7/0018 , G06T7/80 , G06T2207/10004 , H04N5/23216 , H04N5/23293 , H04N17/002
Abstract: A method for dynamically calibrating rotational offset in a device includes obtaining an image captured by a camera of the device. Orientation information of the device at the time of image capture may be associated with the image. Pixel data of the image may be analyzed to determine an image orientation angle for the image. A device orientation angle may be determined from the orientation information. A rotational offset, based on the image orientation angle and the device orientation angle, may be determined. The rotational offset is relative to the camera or orientation sensor. A rotational bias may be determined from statistical analysis of numerous rotational offsets from numerous respective images. In some embodiments, various thresholds and predetermined ranges may be used to exclude some rotational offsets from the statistical analysis or to discontinue processing for that image.
Abstract translation: 用于动态校准装置中的旋转偏移的方法包括获得由装置的照相机拍摄的图像。 图像拍摄时设备的方向信息可能与图像相关联。 可以分析图像的像素数据以确定图像的图像取向角度。 可以根据取向信息来确定设备取向角度。 可以确定基于图像取向角度和装置取向角度的旋转偏移。 旋转偏移量相对于相机或方位传感器。 旋转偏差可以根据来自大量各自图像的许多旋转偏移的统计分析来确定。 在一些实施例中，可以使用各种阈值和预定范围来从统计分析中排除一些旋转偏移或者中断对该图像的处理。
-
公开(公告)号：US12175375B2
公开(公告)日：2024-12-24
申请号：US17903991
申请日：2022-09-06
Applicant: Apple Inc.
Inventor： Gaurav Kapoor , Cecile M. Foret , Francesco Rossi , Kit-Man Wan , Umesh S. Vaishampayan , Etienne Belanger , Albert Antony , Alexey Marinichev , Marco Zuliani , Xiaojin Shi
Abstract: The subject technology provides receiving a neural network (NN) model to be executed on a target platform, the NN model including multiple layers that include operations and some of the operations being executable on multiple processors of the target platform. The subject technology further sorts the operations from the multiple layers in a particular order based at least in part on grouping the operations that are executable by a particular processor of the multiple processors. The subject technology determines, based at least in part on a cost of transferring the operations between the multiple processors, an assignment of one of the multiple processors for each of the sorted operations of each of the layers in a manner that minimizes a total cost of executing the operations. Further, for each layer of the NN model, the subject technology includes an annotation to indicate the processor assigned for each of the operations.
-
-
-
-
-
-
-
-
-
Search Results
39
records
(took 0.024 seconds)
