公开(公告)号：US11157814B2

公开(公告)日：2021-10-26

申请号：US15707064

申请日：2017-09-18

Applicant: Google Inc.

Inventor： Andrew Gerald Howard ,  Bo Chen ,  Dmitry Kalenichenko ,  Tobias Christoph Weyand ,  Menglong Zhu ,  Marco Andreetto ,  Weijun Wang

IPC: G06N3/08 ,  G06N3/04

Abstract: The present disclosure provides systems and methods to reduce computational costs associated with convolutional neural networks. In addition, the present disclosure provides a class of efficient models termed “MobileNets” for mobile and embedded vision applications. MobileNets are based on a straight-forward architecture that uses depthwise separable convolutions to build light weight deep neural networks. The present disclosure further provides two global hyper-parameters that efficiently trade-off between latency and accuracy. These hyper-parameters allow the entity building the model to select the appropriately sized model for the particular application based on the constraints of the problem. MobileNets and associated computational cost reduction techniques are effective across a wide range of applications and use cases.

公开(公告)号：US20180137406A1

公开(公告)日：2018-05-17

申请号：US15707064

申请日：2017-09-18

Applicant: Google Inc.

Inventor： Andrew Gerald Howard ,  Bo Chen ,  Dmitry Kalenichenko ,  Tobias Christoph Weyand ,  Menglong Zhu ,  Marco Andreetto ,  Weijun Wang

IPC: G06N3/04 ,  G06N3/08 ,  G06T7/32

CPC classification number: G06N3/04 ,  G06N3/0454 ,  G06N3/082 ,  G06N3/084 ,  G06T7/32 ,  G06T2207/20081 ,  G06T2207/20084

Abstract: The present disclosure provides systems and methods to reduce computational costs associated with convolutional neural networks. In addition, the present disclosure provides a class of efficient models termed “MobileNets” for mobile and embedded vision applications. MobileNets are based on a straight-forward architecture that uses depthwise separable convolutions to build light weight deep neural networks. The present disclosure further provides two global hyper-parameters that efficiently trade-off between latency and accuracy. These hyper-parameters allow the entity building the model to select the appropriately sized model for the particular application based on the constraints of the problem. MobileNets and associated computational cost reduction techniques are effective across a wide range of applications and use cases.

公开(公告)号：US09069794B1

公开(公告)日：2015-06-30

申请号：US14052157

申请日：2013-10-11

Applicant: Google Inc.

Inventor： Farhana Bandukwala ,  Hartwig Adam ,  John Flynn ,  Krzysztof Sikora ,  Marco Andreetto ,  Arcot J. Preetham

IPC: G06F17/30 ,  G06K9/46

CPC classification number: G06F17/30256 ,  G06F17/30265 ,  G06K9/46

Abstract: Methods, systems, and apparatus are provided for determining location information for images. In one aspect, a method includes obtaining landmark location data from content depicted in an image and corresponding confidence scores. Also, the method includes obtaining caption location data from user input and corresponding confidence scores, and obtaining metadata location data from data provided by an image capturing device. Further, the method includes identifying location pairs from the landmark, caption, and metadata location data, and generating, for each location pair, a geographic consistency score. Additionally, the method includes selecting a location pair based on the geographic consistency scores, and selecting an image location for the image from the selected location pair. Moreover, the method includes determining an image location score based on a confidence score for one of the locations in the selected location pair, and associating the image location and image location score with the image.

Abstract translation: 提供了用于确定图像的位置信息的方法，系统和装置。 一方面，一种方法包括从图像中描绘的内容和相应的置信度得分获得界标位置数据。 此外，该方法包括从用户输入获得字幕位置数据和相应的置信度分数，以及从由图像捕获设备提供的数据获得元数据位置数据。 此外，该方法包括从地标，标题和元数据位置数据识别位置对，并且针对每个位置对，生成地理一致性分数。 另外，该方法包括基于地理一致性分数来选择位置对，以及从所选位置对中选择图像的图像位置。 此外，该方法包括基于所选位置对中的一个位置的置信度分数来确定图像位置分数，以及将图像位置和图像位置分数与图像相关联。