Microsoft Technology Licensing, LLC

Sri International

Sarnoff Corporation

Ying Shan

Zhengyou Zhang

Harpreet Singh Sawhney

Rakesh Kumar

Zicheng Liu

Ying Li

Feng Han

Xu Bin Liu

Jian Jiao

Jianchang Mao

Dong Yu

Gang Hua

Cha Zhang

Charles T Loop

Bogdan Calin Mihai Matei

Steven A N Shafer

Xu Liu

Yanlin Guo

Ying Wu

Steve Hsu

Yue Zhou

Thomas Ryan Hoens

Ryan Cekander

Haijing Wang

Jie Zhu

Jc Mao

A computational search method for retrieving computer information related to a query includes transforming a plurality of candidate answers to candidate answer recurrent binary embedding (RBE) embeddings using a trained RBE model. A query is transformed to a query RBE embedding using the trained RBE model. The query RBE embedding is compared to each candidate answer RBE embedding of a plurality of candidate answer RBE embeddings using a similarity function. The candidate answers are sorted based on the comparisons made using the similarity function, and returning a plurality of the top candidate answers.

Recurrent binary embedding for information retrieval

Systems and methods for providing a predictive framework are provided. The predictive framework comprises plural neural layers of adaptable, executable neurons. Neurons accept one or more input signals and produce an output signal that may be used by an upper-level neural layer. Input signals are received by an encoding neural layer, where there is a 1:1 correspondence between an input signal and an encoding neuron. Input signals for a set of data are received at the encoding layer and processed successively by the plurality of neural layers. An objective function utilizes the output signals of the topmost neural layer to generate predictive results for the data set according to an objective. In one embodiment, the objective is to determine the likelihood of user interaction with regard to a specific item of content in a set of search results, or the likelihood of user interaction with regard to any item of content in a set of search results.

Automated predictive modeling and framework

Functionality is described herein for generating a model on the basis of user-behavioral data and knowledge data. In one case, the user-behavioral data identifies queries submitted by users, together with selections made by the users in response to the queries. The knowledge data represents relationships among linguistic items, as expressed by one or more structured knowledge resources. The functionality leverages the knowledge data to supply information regarding semantic relationships which may not be adequately captured by the user-behavioral data, to thereby produce a more robust and accurate model (compared to a model produced on the basis of only user-behavioral data). Functionality is also described herein for applying the model, once trained. In one case, the model may correspond to a deep learning model.

Generating and using a knowledge-enhanced model

Videos are retargeted to a target display for viewing with little to no geometric distortion or video information loss. Salient regions of video frames may be determined using scale-space spatiotemporal information. Video information loss may be a result of spatial loss, due to cropping, and resolution loss, due to resizing. A desired cropping window may be determined using a coarse-to-fine searching strategy. Video frames may be cropped with a window that matches an aspect ratio of the target display, and resized isotropically to match a size of the target display.

Video retargeting

Computer-readable media and computerized methods for utilizing an awareness of video content within a media file to incorporate an advertisement therein are provided. Building the video-content awareness involves performing an offline authoring process to generate a trajectory. In particular, the offline authoring process includes targeting an object appearing in the video content, tracking movement of the object over a sequence of frames within the media file, and, based on the tracked movement, writing locations of the object to the trajectory. In an online rendering process, invoked upon initiating play of the media file, the advertisement is automatically selected and dynamically placed on top of the video content as a function of the locations within the trajectory. Accordingly, by repositioning the advertisement (typically an animation) within the video content based on the movement of the object, an illusion that the advertisement visually interacts with the object is generated.

Video content-aware advertisement placement

Methods and computer-readable media for presenting a user with advertising content, such as an advertisement or coupon, based on the user's responses to one or more questions are provided. A user may view a web page that includes one or more thumbnails, where each thumbnail represents a media, such as a video clip, an audio clip, or an image. Once a user selects a thumbnail, by hovering a cursor over the thumbnail, for example, a question may appear over the thumbnail. The user may respond to the question, and based on the response, may be provided with some form of relevant advertising content. Subsequent questions may be presented to the user if the user selects more than one thumbnail in a session, and thus the user may be provided with highly relevant advertising content that is based on responses to more than one question.

Active and progressive targeted advertising

Methods and computer storage media for creating contextual-display advertisements are provided. A request for a contextual-display advertisement to be presented on a page is received. Upon receiving the request, the page is evaluated to determine the context of the page so that the resulting advertisement is contextually related to the requesting page. The contextually related text and base image are extracted from their respective database to be merged into a contextual-display advertisement. The visual characteristics of the advertisement text are manipulated to provide visual congruity among the elements of the advertisement. Additionally, the placement of the text on the base image is optimized to provide a contextual-image advertisement that is then rendered to be presented on the requesting page.

Contextual-display advertisement

The present invention provides a computer implemented process for detecting multi-view multi-pose objects. The process comprises training of a classifier for each intra-class exemplar, training of a strong classifier and combining the individual exemplar-based classifiers with a single objective function. This function is optimized using the two nested AdaBoost loops. The first loop is the outer loop that selects discriminative candidate exemplars. The second loop, the inner loop selects the discriminative candidate features on the selected exemplars to compute all weak classifiers for a specific position such as a view/pose. Then all the computed weak classifiers are automatically combined into a final classifier (strong classifier) which is the object to be detected.

System and method for detection of multi-view/multi-pose objects

A method and apparatus for recognizing an object, comprising providing a set of scene features from a scene, pruning a set of model features, generating a set of hypotheses associated with the pruned set of model features for the set of scene features, pruning the set of hypotheses, and verifying the set of pruned hypotheses is provided.

Method and apparatus for recognizing 3-D objects

Systems, methods, computer-readable media, and graphical user interfaces for facilitating advertisement placement over video content are provided. Images within a video are partitioned into image regions. Upon partitioning images into image regions, an intrusiveness score is determined for each image region. Based on the intrusiveness scores, optimal placement of an advertisement within the video is determined.

Facilitating advertisement placement over video content

A method and apparatus for tracking a movable object using a plurality of images, each of which is separated by an interval of time is disclosed. The plurality of images includes first and second images. The method and apparatus include elements for aligning the first and second images as a function of (i) at least one feature of a first movable object captured in the first image, and (ii) at least one feature of a second movable object captured in the second image; and after aligning the first and second images, comparing at least one portion of the first image with at least one portion of the second image.

Method and apparatus for tracking a movable object

The present invention provides an improved system and method for object detection with histogram of oriented gradient (HOG) based support vector machine (SVM). Specifically, the system provides a computational framework to stably detect still or not moving objects over a wide range of viewpoints. The framework includes providing a sensor input of images which are received by the 'focus of attention' mechanism to identify the regions in the image that potentially contain the target objects. These regions are further computed to generate hypothesized objects, specifically generating selected regions containing the target object hypothesis with respect to their positions. Thereafter, these selected regions are verified by an extended HOG-based SVM classifier to generate the detected objects.

System and method for detecting still objects in images

A method and apparatus for unsupervised learning of measures for matching objects between images from at least two non-overlapping cameras is disclosed The method includes collecting at least two pairs of feature maps, where the at least two pairs of feature maps are derived from features of objects captured in the images. The method further includes computing, as a function of at least two pairs of feature maps, at least one first and second match measures, wherein the first match measure is of a same class and the second match measure is of a different class.

Method and apparatus for unsupervised learning of discriminative edge measures for vehicle matching between non-overlapping cameras

A system and method for identifying objects, particularly vehicles, between two non-overlapping cameras. More specifically, the method and system determines whether a vehicle depicted in an image captured by a first camera is the same vehicle or a different vehicle than a vehicle depicted in an image captured by a second camera. This intra-camera analysis determines whether the vehicle viewed by the first camera is the same as the vehicle viewed by the second camera, without directly matching the two vehicle images, thus eliminating the problems and inaccuracies caused by disparate environmental conditions acting on the two cameras, such as dramatic appearance and aspect changes.

Growing community of inventors

Sammamish, WA, United States of America

Ying Shan