Abstract: This disclosure describes a method and system to perform object detection and 6D pose estimation. The system comprises a database of 3D models, a CNN-based object detector, multiview pose verification, and a hard example generator for CNN training. The accuracy of that detection and estimation can be iteratively improved by retraining the CNN with increasingly hard ground truth examples. The additional images are detected and annotated by an automatic process of pose estimation and verification.

Abstract: This disclosure describes a method and system to perform object detection and 6D pose estimation. The system comprises a database of 3D models, a CNN-based object detector, multiview pose verification, and a hard example generator for CNN training. The accuracy of that detection and estimation can be iteratively improved by retraining the CNN with increasingly hard ground truth examples. The additional images are detected and annotated by an automatic process of pose estimation and verification.

Abstract: Techniques are provided for dynamically adapting the view from a conference endpoint that includes a presentation apparatus, such as a whiteboard. A first signal is received that includes a video signal derived from a video camera that is viewing a room during a conference session in which a person is presenting information on a presentation apparatus. During the video conference, switching is performed between the first signal and a second signal representing content being displayed on the presentation apparatus during the conference session for output and transmission to other conference endpoints of the conference session. The determination as to whether to supply the first signal (for a normal view of the conference room) or the second signal may be based on a position determination of the presenter or may be instead be based on an external view selection command received from another conference endpoint participating in the conference session.

Abstract: Techniques are provided for establishing a videoconference session between participants at different endpoints, where each endpoint includes at least one computing device and one or more displays. A plurality of video streams is received at an endpoint, and each video stream is classified as at least one of a people view and a data view. The classified views are analyzed to determine one or more regions of interest for each of the classified views, where at least one region of interest has a size smaller than a size of the classified view. Synthesized views of at least some of the video streams are generated, wherein the synthesized views include at least one view including a region of interest, and views including the synthesized views are rendered at one or more displays of an endpoint device.

Abstract: A method is provided in one example and includes receiving audio data at a microphone array that includes a plurality of microphones. The microphone array is provisioned at a first endpoint, which includes a camera element configured to capture video data associated with a video session involving the first endpoint and a second endpoint. The method also includes formatting the audio data into a time division multiplex (TDM) stream, and communicating the stream to a port for a subsequent communication over a network and to the second endpoint.

Abstract: An apparatus is provided in one example and includes a camera configured to receive image data associated with an end user involved in a video session. The apparatus also includes a display and an optics element configured to interface with the camera. The optics element reflects the image data associated with the end user positioned in front of the display. A retracting mechanism is also provided and is configured to retract the optics element in a direction such that the camera moves to an inactive state and the optics element is removed from a view of the display from the perspective of the end user. An effective optical distance from the camera to the end user is increased by manipulating a position of the optics element. In more detailed embodiments, the camera can be configured above the display such that its lens points downward toward the optics element.

Abstract: In one embodiment, a method includes receiving audio at a plurality of microphones, identifying a sound source to be muted, processing the audio to remove sound received from the sound source at each of the microphones, and transmitting the processed audio. An apparatus is also disclosed.

Abstract: A method is provided in one example and includes receiving image data for a field of view associated with a display. The image data is used to generate a plurality of red green blue (RGB) frames. The method also includes emitting infrared energy onto the field of view in order to generate a plurality of infrared frames, the plurality of RGB frames and the plurality of infrared frames are generated by a single camera. The plurality of RGB frames can be combined with the plurality of infrared frames in order to generate a video data stream. In a more particular embodiment, the emitting of the infrared energy is synchronized with the camera such that the infrared energy is emitted onto the field of view at one half of an existing frame rate of the camera.

Abstract: A method for transition control in a videoconference comprises receiving a plurality of video streams from a plurality of cameras, displaying a first video stream of the plurality of video streams, detecting a stream selection event for display of a second video stream of the plurality of video streams, determining a transition category for a transition from the first video stream to the second video stream, and selecting a display transition based on the transition category for displaying the transition from the first video stream to the second video stream.

Abstract: Techniques are provided for dynamically adapting the view from a conference endpoint that includes a presentation apparatus, such as a whiteboard. A first signal is received that includes a video signal derived from a video camera that is viewing a room during a conference session in which a person is presenting information on a presentation apparatus. During the video conference, switching is performed between the first signal and a second signal representing content being displayed on the presentation apparatus during the conference session for output and transmission to other conference endpoints of the conference session. The determination as to whether to supply the first signal (for a normal view of the conference room) or the second signal may be based on a position determination of the presenter or may be instead be based on an external view selection command received from another conference endpoint participating in the conference session.

Abstract: Clock synchronization for an acoustic echo canceller (AEC) with a speaker and a microphone connected over a digital link may be provided. A clock difference may be estimated by analyzing the speaker signal and the microphone signal in the digital domain. The clock synchronization may be combined in both hardware and software. This synchronization may be performed in two stages, first with coarse synchronization in hardware, then fine synchronization in software with, for example, a re-sampler.

Abstract: Techniques are provided for dynamically adapting the view from a conference endpoint that includes a presentation apparatus, such as a whiteboard. A first signal is received that includes a video signal derived from a video camera that is viewing a room during a conference session in which a person is presenting information on a presentation apparatus. During the video conference, switching is performed between the first signal and a second signal representing content being displayed on the presentation apparatus during the conference session for output and transmission to other conference endpoints of the conference session. The determination as to whether to supply the first signal (for a normal view of the conference room) or the second signal may be based on a position determination of the presenter or may be instead be based on an external view selection command received from another conference endpoint participating in the conference session.

Abstract: Techniques are provided for establishing a videoconference session between participants at different endpoints, where each endpoint includes at least one computing device and one or more displays. A plurality of video streams is received at an endpoint, and each video stream is classified as at least one of a people view and a data view. The classified views are analyzed to determine one or more regions of interest for each of the classified views, where at least one region of interest has a size smaller than a size of the classified view. Synthesized views of at least some of the video streams are generated, wherein the synthesized views include at least one view including a region of interest, and views including the synthesized views are rendered at one or more displays of an endpoint device.

Abstract: A method is provided in one example and includes capturing panoramic image data through a first camera in a camera cluster, and capturing close-up image data through a second camera included as part of a spaced array of cameras. The presence of a user in a field of view of the second camera can be detected. The close-up image data and the panoramic image data can be combined to form a combined image. In more specific embodiments, the detecting includes evaluating a distance between the user and the second camera. The combined image can reflect a removal of a portion of panoramic image data associated with the user in a video conferencing environment.

Abstract: Techniques are provided for establishing a videoconference session between participants at different endpoints, where each endpoint includes at least one computing device and one or more displays. A plurality of video streams is received at an endpoint, and each video stream is classified as at least one of a people view and a data view. The classified views are analyzed to determine one or more regions of interest for each of the classified views, where at least one region of interest has a size smaller than a size of the classified view. Synthesized views of at least some of the video streams are generated, wherein the synthesized views include at least one view including a region of interest, and views including the synthesized views are rendered at one or more displays of an endpoint device.

Abstract: Particular embodiments generally relate to video compression. In one embodiment, a store of reference frames is provided in memory. The reference frames may be classified based on a plurality of classifiers. The classifiers may correspond to features that are found in the reference frame. A frame to encode is then received. The frame is analyzed to determine features found in the frame. As macroblocks in the frame are encoded, a macroblock is analyzed to determine which feature may be included in the macroblock. The feature is used to determine a classifier, which is used to determine a subset of the reference frames. The subset is then searched to determine a reference frame for the macroblock.

Abstract: Video frames are captured at one or more cameras during a video conference session, where each video frame includes a digital image with a plurality of pixels. Depth values associated with each pixel are determined in at least one video frame, where each depth value represents a distance of a portion of the digital image represented by at least one corresponding pixel from the one or more cameras that capture the at least one video frame. Luminance values of pixels are adjusted within captured video frames based upon the depth values determined for the pixels so as to achieve relighting of the video frames as the video frames are displayed during the video conference session.

Abstract: Techniques are described herein for combining video frames of two or more real-time video streams into combined video frames of a combined real-time video stream for continuous presence. Video frames of at least two real-time video streams are combined into combined video frames of a combined video stream. The combined video stream is supplied to a video display for displaying the combined video stream. Each video stream includes video frames with subject and background images. The subject images of corresponding video frames of the first and second video streams are combined into a combined video frame of a combined video stream such that the subject image of the first video stream is positioned in an anterior portion of the combined frame and the subject image of the second video stream is positioned in a posterior portion of the combined frame.

Abstract: A method that includes capturing depth information associated with a first field of view of a depth camera. The depth information is represented by a first plurality of depth pixels. The method also includes capturing color information associated with a second field of view of a video camera that substantially overlaps with the first field of view of the depth camera. The color information is represented by a second plurality of color pixels. The method further includes enhancing color information represented by at least one color pixel of the second plurality of color pixels to generate an enhanced image. The enhanced image adjusts an exposure characteristic of the color information captured by the video camera. The at least one color pixel is enhanced based on depth information represented by at least one corresponding depth pixel of the first plurality of depth pixels.

Abstract: Systems and methods for providing camera configuration for points in a multi-point videoconference system are provided. First configuration information is determined for a first point of a multi-point videoconferencing system. Second configuration information is determined for a second point of the multi-point videoconferencing system. One or more first cameras at the first point or one or more second cameras at the second point of the multi-point videoconferencing system are reconfigured based on the first configuration information or the second configuration information.