Google Inc.

Amazon Technologies, Inc.

Ming Zhao

Yang Song

Hartwig Adam

Jay Yagnik

Henry Allan Rowley

Yushi Jing

Ullas Gargi

Tomas Isdal

Tze Way Eugene Ie

Sherif M Yacoub

Fei Sha

Hexiang Hu

Joonseok Lee

Vihan Jain

Sheide Chammas

Bowen Zhang

Michael Patrick Bacus

Hong Chen

Xuping Zhang

Xi Long

Bingbing Ni

Yantao Zheng

Sanjiv Kumar

Gideon S Mann

Corinna Cortes

Ameesh Makadia

Hanning Zhou

Jian Lin

Junqing Shang

Manikandan Gopalakrishnan

Hartwig Adams

Zheshen Wang

A computer-implemented method for generating video representations utilizing a hierarchical video encoder includes obtaining a video, wherein the video includes a plurality of frames, processing each of the plurality of frames with a machine-learned frame-level encoder model to respectively generate a plurality of frame representations for the plurality of frames, the plurality of frame representations respective to the plurality of frames determining a plurality of segment representations representative of a plurality of video segments including one or more of the plurality of frames, the plurality of segment representations based at least in part on the plurality of frame representations, processing the plurality of segment representations with a machine-learned segment-level encoder model to generate a plurality of contextualized segment representations, determining a video representation based at least in part on the plurality of contextualized segment representations, and providing the video representation as an output.

Hierarchical video encoders

A computer-implemented method for executing a software application in a virtual machine executing on a data processing device includes receiving software code for a software application, determining application programming interfaces referenced by the software code, determining portions of an operating system accessed by the software code and creating an application container in the virtual machine. The method also includes application programming interfaces referenced by the software code inside the application container, portions of the operating system accessed by the software code inside the application container and executing the software application inside the application container on the virtual machine.

Virtual machine

A method includes executing a virtual machine, which executes on a data processing device and includes an application container and an agent. The application container includes a file system mount interfaced with a file system residing outside the application container, an application programming interface (API), and a software application including a reference to the API. The agent is configured to monitor whether the software application attempts to access any resources outside of the application container. When the software application attempts to access any resources outside of the application container, the agent is configured to execute a remediation routine.

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for associating still images and videos. One method includes receiving a plurality of images and a plurality of videos and determining whether the images are related to the videos. The determining includes, for an image and a video, extracting features from the image and extracting features frames of the video, and comparing the features to determine whether the image is related to the video. The method further includes maintaining a data store storing data associating each image with each video determined to be related to the image.

Associating still images and videos

A document analysis module analyzes electronic media items and identifies regions and region types for the electronic media items. The document analysis module may use rules, typographical feature sets, and cluster analysis to identify regions and region types. The document analysis module may also receive user input and may use the user input to identify regions and region types. The document analysis module may further use template pages to identify regions and region types.

Document analysis for region classification

Identifying responsive resources across still images and videos

A distributed system including one or more data processing devices executing instructions configuring the one or more data processing device to execute a virtual machine including a software application and an agent. The agent is configured to receive a first health status of the software application, determine a second health status of the software application, compare the first health status and the second health status, and restart the software application based on the comparison. The one or more data processing device also execute an application server configured to remotely determine the first health status of the software application and send the first health status of the software application to the virtual machine.

A computing device hosting a server identifies one or more potential typographical errors in a work using an initial reference. The computing device sends data indicative of the presence of the potential typographical errors to users. The computing device collects feedback for the work from the users. The feedback for the work indicates whether the potential typographical errors are author-intended strings. The computing device combines the user feedback for the work with user feedback for other works and sorts the combined user feedback based on one or more selected parameters. The computing device determines, based on the sorted user feedback, that one or more of the potential typographical errors includes an acceptable string and updates the initial reference to include the acceptable string.

Classification of a string as a typographical error based on feedback

In some implementations, legibility of an image may be automatically determined based, at least in part, on text contained in the image. For example, image analysis techniques may be used to identify text components in an image. One or more features of each text component may be determined for use in assessing the legibility of the text component. For example, a classifier trained on the one or more features may provide a confidence level indicative of the legibility of each text component. The confidence level for each of the text components may be compared to a legibility threshold for determining whether the text component is legible or illegible. Based, at least in part, on the determination as to how much of the text in the image is legible or illegible, an overall legibility of the image may be assessed.

Assessing legibility of images

This invention relates to building a landmark database from web data. In one embodiment, a computer-implemented method builds a landmark database. Web data including a web page is received from one or more websites via one or more networks. The web data is interpreted using at least one processor to determine landmark data describing a landmark. At least a portion of the landmark data identifies a landmark. Finally, a visual model is generated using the landmark data. A computing device is able to recognize the landmark in an image based on the visual model.

Web mining to build a landmark database and applications thereof

Growing community of inventors

Bellevue, WA, United States of America

Ming Zhao