We present an algorithm for simultaneous face detection, landmarks localization, pose estimation and gender recognition using deep convolutional neural networks (CNN). The proposed method called, HyperFace, fuses the intermediate layers of a deep CNN using a separate CNN followed by a multi-task learning algorithm that operates on the fused features. It exploits the synergy among the tasks which boosts up their individual performances. Additionally, we propose two variants of HyperFace: (1) HyperFace-ResNet that builds on the ResNet-101 model and achieves significant improvement in performance, and (2) Fast-HyperFace that uses a high recall fast face detector for generating region proposals to improve the speed of the algorithm. Extensive experiments show that the proposed models are able to capture both global and local information in faces and performs significantly better than many competitive algorithms for each of these four tasks.

HyperFace: A Deep Multi-Task Learning Framework for Face Detection, Landmark Localization, Pose Estimation, and Gender Recognition

A mobile terminal and a control method thereof, which can obtain an image, are provided. A mobile terminal (100) includes a camera unit (121), a pictogram extraction unit (182) and a controller (180). The camera unit (121) obtains image information corresponding to at least one of a still image and a moving image. The pictogram extraction unit (182) extracts at least one pictogram from the obtained image information. The controller (180) detects information related to the extracted pictogram, and displays the detected information to be overlapped with the obtained image information. In the mobile terminal, the controller (180) includes, as the detected information, at least one of previously recorded information and currently searched information related to the extracted pictogram.

Mobile terminal and control method thereof

A media device allows users to watch and capture portions from a media stream. Users may then share the captured media content with other users. In one embodiment, the media device receives a media stream, plays the media stream, and caches a portion of the media stream as it is being played. A user can define a media clip by selecting its boundaries in the cached portion of the media stream. The media device creates the media clip based on the user's input and enables the user to transmit the media clip to another system, such as a community website for sharing it with other users.

Capturing and sharing media content

A computer-implemented method includes identifying a computer-based portable program module, automatically altering code in the portable program module to permit display of the module on a television-based display so that the displayed module has a substantially similar appearance on the television-based display as on a computer display, and providing the altered code for execution on a processor connected to a television-based display.

Interactive Media Display Across Devices

Recommendation engine apparatus and associated methods provide content compiled from various sources and selected to match user preferences. In one embodiment, the recommendation apparatus comprises a headend entity; in another, it is co-located on a user's CPE. In one embodiment, the recommendation engine creates content records from content metadata for comparison to a user profile. The user profile is pre-programmed; however has the ability to dynamically shift toward a user's preferences as the user takes actions regarding content. Client applications are utilized to compile and present content; feedback mechanisms are utilized to enable “learning” from user activities to generate more precise recommendations as well as to “unlearn” stale preferences. Recommended content is displayed in the form of a playlist, or as a continuous stream on a virtual channel, or presented in an electronic program guide. A business rules “engine” useful in implementing operational or business goals is also disclosed.

Recommendation engine apparatus and methods

The invention relates to systems and methods for providing video segments over a network. In one embodiment, metadata identifying a video segment is received from a first client device and transmitted to a second client device on a platform different than the first client device, via a network. In another embodiment, metadata identifies a key frame of a video file, which may be used to retrieve and/or playback a portion of the video file via the network. In another embodiment, metadata, generated at a plurality of client devices that are linked via the network, are indexed.

Systems and methods of searching for and presenting video and audio

The invention relates to methods and apparatus for providing media assets over a network. First metadata corresponding to a first video asset is generated. The first metadata includes text describing contents displayed when the first video asset is played and a pointer to a location within a video file that corresponds to the first video asset. The pointer includes at least two of a start location, an end location, and a duration. The first metadata is transmitted for receipt by a client system capable of playing the first video asset. The client system displays portions of the text of the first metadata to a user of the client system, and uses the pointer of the first metadata to facilitate requesting the first video asset from a video server for transmitting video assets over the network.

Methods and systems for providing media assets over a network

Systems and methods for generating a composite video having a plurality of video assets are provided. Systems may include storage having a plurality of video assets, a metadata generator, and a composite video generator. The metadata generator processes respective ones of the video assets to generate a metadata track representative of information that is descriptive of the content of the video asset. The composite video generator receives a plurality of metadata tracks and, in response, processes the associated video assets and the metadata tracks to generate a composite video asset having video data from the video assets. The composite video asset may have a metadata list panel that presents information representative of the metadata information as video data appearing within the composite video asset. The metadata list panel visually presents the sequence of video assets in the composite video asset.

Mark Pascarella

Papers

Systems and methods of searching for and presenting video and audio

Methods and systems for providing media assets over a network

Methods and systems for generating and delivering navigatable composite videos