Amazon Technologies, Inc.

Cisco Technology, Inc.

Twitter, Inc.

Mark Julio Kalman

Bin Wang

Geraint John Davies

Vinod Murli Mamtani

Yon-Seo Kim

Gerard Joseph Heinz, Ii

Scott Wright Heath

Keith Emery Belovay

Implementations provide low-latency live-video streams using existing content delivery networks. An example method includes receiving a video broadcast as a series of frames and determining, for each frame, whether the frame is a break frame. Responsive to determining that the frame is a break frame, the method includes removing an in-progress tag from a current segment file in a playlist for the video broadcast. The playlist includes at least a previous segment file, the current segment file, and a next segment file, which also has a respective in-progress tag. The method also includes associating the frame with a next segment file in a playlist and transmitting the playlist to a cache server. Responsive to determining the frame in the series of frames is not a break frame, the method includes associating the frame with the current segment file. The frame is transmitted to the cache server as a chunk.

Low-latency HTTP live streaming

Implementations provide a structure for low-latency live-video streams using existing content delivery networks. An example method includes creating, at the start of a broadcast stream, a playlist for the broadcast stream that identifies at least two segment files. The content of each of the segment files is a table header. The method also includes transmitting the playlist and each of the segment files to a cache server, obtaining the video broadcast stream as a series of frames, and transmitting each frame in the series of frames to the cache server using an HTTP chunk transfer protocol, each frame being associated with, and written to, a first segment file identified in the playlist. When a frame in the broadcast stream is a break frame, the method includes transmitting an end of file marker for the first segment file and adding a video header for the break frame before it is transmitted.

Techniques for efficient bandwidth estimation are described herein. In some cases, the bandwidth estimation techniques disclosed herein may, for example, calculate bandwidth based on multiple packet groups transmitted at different times. Additionally, in some cases, the bandwidth estimation techniques disclosed herein may, for example, capture cross traffic and its effects on bandwidth. Furthermore, in some cases, the bandwidth estimation techniques disclosed herein may, for example, employ dynamic self-correcting techniques for more reliable estimates.

Efficient bandwidth estimation

Techniques for streaming content, such as a video game or whiteboard application, to multiple different clients are described herein. In some cases, differences may exist between characteristics of the different clients such as screen size, resolution, operating system, processors and memories and others. The different clients may be positioned in close proximity to one another or at remote locations with respect to one another. In some cases, different network conditions, such as available bandwidth, loss rates, latency, throughput and others, may exist for transmitting content to the different clients.

Streaming content to multiple clients

Techniques for adaptive content transmission are described herein. During transmission of a content item, a network connection may be monitored to collect data corresponding to one or more network conditions associated with the transmission of the content item. Such network conditions may include, for example, network throughput, available network bandwidth, network latency and others. The collected data may be used to dynamically adjust one or more transmission attributes in connection with the transmitted content item. The one or more transmission attributes may be determined for adjustment at any desired transmission interval.

Adaptive content transmission

A method for generating an interpolated output image, including generating a first interpolation array including RGB values, at least one of which is interpolated, based on a first interpolation region, and generating a first homogeneity array based on the first interpolation array including homogeneity values calculated based on a first homogeneity neighborhood having the same pixel locations as the first interpolation region. The method includes generating a second interpolation array based on a second interpolation region, and generating a second homogeneity array based on the second interpolation array and a second homogeneity neighborhood, where the second homogeneity neighborhood includes the same pixel locations as the second interpolation region. The method includes determining, for a first pixel location, that the homogeneity value in the first homogeneity array is greater than the homogeneity value in the second homogeneity array, and outputting the RGB color values in the first interpolation array.

System and method for image demosaicing

Embodiments of the invention provide a technique to sharpen luminance data associated with each pixel of a digital image without introducing certain sharpening artifacts. An alpha mask is generated based on first, second, and third alpha submasks that indicate how strongly overshoot, color fringe, and dark regions, respectively, are detected at each pixel location. The alpha mask is used to combine luminance data derived from the digital image with sharpened luminance data that is generated by applying a sharpening filter to the luminance data.

Growing community of inventors

Alameda, CA, United States of America

Mark Julio Kalman