Abstract: In an example method, a decoder obtains a data stream representing video content. The video content is partitioned into one or more logical units, and each of the logical units is partitioned into one or more respective logical sub-units. The decoder determines that the data stream includes first data indicating that a first logical unit has been encoded according to a flexible skip coding scheme. In response, the decoder determines a first set of decoding parameters based on the first data, and decodes each of the logical sub-units of the first logical unit according to the first set of decoding parameters.

Abstract: A cross-component based filtering system is disclosed for video coders and decoders. The filtering system may include a filter having an input for a filter offset and an input for samples reconstructed from coded video data representing a native component of source video on which the filter operates. The offset may be generated at least in part from a sample classifier that classifies samples reconstructed from coded video data representing a color component of the source video orthogonal to the native component according to sample intensity.

Abstract: Techniques are disclosed for generating virtual reference frames that may be used for prediction of input video frames. The virtual reference frames may be derived from already-coded reference frames and thereby incur reduced signaling overhead. Moreover, signaling of virtual reference frames may be avoided until an encoder selects the virtual reference frame as a prediction reference for a current frame. In this manner, the techniques proposed herein contribute to improved coding efficiencies.

Abstract: Video coders and decoders perform transform coding and decoding on blocks of video content according to an adaptively selected transform type. The transform types are organized into a hierarchy of transform sets where each transform set includes a respective number of transforms and each higher-level transform set includes the transforms of each lower-level transform set within the hierarchy. The video coders and video decoders may exchange signaling that establishes a transform set context from which a transform set that was selected for coding given block(s) may be identified. The video coders and video decoders may exchange signaling that establishes a transform decoding context from which a transform that was selected from the identified transform set to be used for decoding the transform unit. The block(s) may be coded and decoded by the selected transform.

Abstract: Techniques are disclosed for deriving prediction pixel blocks for use in intra-coding video and combined inter- and intra-coding video. In a first aspect, the techniques may include deriving value(s) for pixel location(s) of the prediction pixel block by, when a prediction direction vector assigned to the prediction vector points to quadrants I or III of a Cartesian plane, deriving the pixel location's value from pixel values in two regions of previously-decoded pixel data intercepted by extending the prediction direction vector in two opposite directions through the pixel location. When the prediction direction vector points toward quadrants II of the Cartesian plane, deriving the pixel location's value from pixel values in one region intercepted by the prediction direction vector through the pixel location, and from a second region intercepted by a vector that is orthogonal to the prediction direction vector.

Abstract: Techniques are disclosed for coding video in applications where regions of video are inactive on a frame to frame basis. According to the techniques, coding processes update and reconstruct only a subset of pixel blocks of pixels within a frame, while other pixel blocks are retained from a previously coded frame stored in a coder's or decoder's reference frame buffer. The technique is called Backward Reference Updating (or “BRU”) for convenience. At a desired pixel block granularity, based on the activity between a current frame to be coded and its reference frame(s), BRU will only perform prediction, transform, quantization, and reconstruction on selected regions that are determined to be active. The reconstructed pixels in these active regions are directly placed onto a specified reference frame in memory instead of creating a new frame. Therefore, fewer memory transfers need to be performed.

Abstract: Systems and methods are configured for accessing data representing video content, the data comprising a set of one or more symbols each associated with a syntax element; performing a probability estimation, for encoding the data, comprising: for each symbol, obtaining, based on the syntax element for that symbol, an adaptivity rate parameter value, the adaptivity rate parameter value being a function of a number of symbols in the set of one or more symbols; updating the adaptivity rate parameter value as a function of an adjustment parameter value; and generating, based on the updated adaptivity rate parameter value, a probability value; generating a probability estimation; and encoding, based on the CDF of the probability estimation, the data comprising the set of one or more symbols for transmission.

Abstract: Disclosed is a method that includes receiving an image frame having a plurality of coded blocks, determining a prediction unit (PU) from the plurality of coded blocks, determining one or more motion compensation units arranged in an array within the PU, and applying a filter to one or more boundaries of the one or more motion compensation units. Also disclosed is a method that includes receiving a reference frame that includes a reference block, determining a timing for deblocking a current block, performing motion compensation on the reference frame to obtain a predicted frame that includes a predicted block, performing reconstruction on the predicted frame to obtain a reconstructed frame that includes a reconstructed PU, and applying, at the timing for deblocking the current block, a deblocking filter based on one or more parameters to the reference block, the predicted block, or the reconstructed PU.

Abstract: Systems and methods disclosed for video compression, utilizing neural networks for predictive video coding. Processes employed combine multiple banks of neural networks with codec system components to carry out the coding and decoding of video data.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for reducing a quantity of buffers for a video codec. One of the methods includes determining, from a plurality of prediction modes, a prediction mode for data that represents frame data in a frame; in response to determining the prediction mode, selecting, using the prediction mode, one or more buffers from a plurality of buffers, each buffer of which is for a prediction mode from the plurality of prediction modes, a first quantity of buffers in the plurality of buffers being less than a second quantity of prediction modes in the plurality of prediction modes; retrieving, from each of the one or more buffers, historical data for the frame data; and in response to retrieving the historical data, generating, using the historical data, updated data for the frame data in the frame of the video sequence.

Abstract: Techniques are disclosed by which a coding parameter is determined to encode video data resulting in encoded video data possessing a highest possible video quality. Features may be extracted from an input video sequence. The extracted features may be compared to features described in a model of coding parameters generated by a machine learning algorithm from reviews of previously-coded videos, extracted features of the previously-coded videos, and coding parameters of the previously-coded videos. When a match is detected between the extracted features of the input video sequence and extracted features represented in the model, a determination may be made as to whether coding parameters that correspond to the matching extracted feature correspond to a tier of service to which the input video sequence is to be coded.

Abstract: A system obtains a data set representing immersive video content for display at a display time, including first data representing the content according to a first level of detail, and second data representing the content according to a second higher level of detail. During one or more first times prior to the display time, the system causes at least a portion of the first data to be stored in a buffer. During one or more second times prior to the display time, the system generates a prediction of a viewport for displaying the content to a user at the display time, identifies a portion of the second data corresponding to the prediction of the viewport, and causes the identified portion of the second data to be stored in the video buffer. At the display time, the system causes the content to be displayed to the user using the video buffer.

Abstract: Techniques are proposed to improve temporal motion projection in video coding. Candidate reference frames available for use in temporal motion projection are sorted in processing order according to scores assigned based on estimates of the reference frames' suitability for prediction. Such estimates may be based on temporal distance between each candidate reference frame and that reference frame's prediction references. Estimates may be based, for each reference frame, based on an estimate of coding quality of a reference frame from which the respective candidate reference frame makes a prediction reference. Once sorted, the candidate reference frames may be processing in the sorting order to supply prediction data to a current frame that is to be coded from the candidate reference frames. Additionally, hardware friendly designs of motion field hole filling and motion vector smoothing operations are proposed.

Abstract: In an example method, a decoder accesses a bitstream representing video content, and parses one or more flexible coefficient position (FCP) syntax from the bitstream, where the one or more FCP syntax indicate one or more index values. The decoder further determines side information representing one or more characteristics of an encoded portion of the video content. The decoder interprets the one or more FCP syntax based on the side information, including determining a coefficient position with respect to the encoded portion of the video content based on the one or more index values and the side information. The decoder decodes the encoded portion of the video content according to the coefficient position.

Abstract: The present disclosure describes techniques for efficient coding of motion vectors developed for multi-hypothesis coding applications. According to these techniques, when coding hypotheses are developed, each having a motion vector identifying a source of prediction for a current pixel block, a motion vector for a first one of the coding hypotheses may be predicted from the motion vector of a second coding hypothesis. The first motion vector may be represented by coding a motion vector residual, which represents a difference between the developed motion vector for the first coding hypothesis and the predicted motion vector for the first coding hypothesis, and outputting the coded residual to a channel. In another embodiment, a motion vector residual may be generated for a motion vector of a first coding hypothesis, and the first motion vector and the motion vector residual may be used to predict a second motion vector and a predicted motion vector residual.

Abstract: A system obtains a data set representing immersive video content for display at a display time, including first data representing the content according to a first level of detail, and second data representing the content according to a second higher level of detail. During one or more first times prior to the display time, the system causes at least a portion of the first data to be stored in a buffer. During one or more second times prior to the display time, the system generates a prediction of a viewport for displaying the content to a user at the display time, identifies a portion of the second data corresponding to the prediction of the viewport, and causes the identified portion of the second data to be stored in the video buffer. At the display time, the system causes the content to be displayed to the user using the video buffer.

Abstract: Improved lossless entropy coding techniques for coding of image data include selecting a context for entropy coding based on an ordered scan path of possible context locations. A symbol for a current location within a source image may be entropy coded based on a context of prior encoded symbols of other locations within source images, where the context is selected based on an ordered scan path enumerating a series of potential context locations within one or more source images. To select a context, a predetermined number of prior symbols may be selected by qualifying or disqualifying locations in the scan path, and then the current symbol may be encoded with a context based on prior symbols corresponding to the first qualifying context locations in the order of the scan path.

Abstract: Techniques for multi-view video streaming are described in the present disclosure, wherein a viewport prediction may be employed at a client-end based on analysis of pre-fetched media item data and ancillary information. A streaming method may first prefetch a portion of content of a multi-view media item. The method may next identify a salient region from the prefetched content and may then download additional content of the media item that corresponds to the identified salient region.

Abstract: A system obtains a data set representing immersive video content for display at a display time, including first data representing the content according to a first level of detail, and second data representing the content according to a second higher level of detail. During one or more first times prior to the display time, the system causes at least a portion of the first data to be stored in a buffer. During one or more second times prior to the display time, the system generates a prediction of a viewport for displaying the content to a user at the display time, identifies a portion of the second data corresponding to the prediction of the viewport, and causes the identified portion of the second data to be stored in the video buffer. At the display time, the system causes the content to be displayed to the user using the video buffer.

Abstract: Techniques are disclosed for improved video coding with virtual reference frames. A motion vector for prediction of a pixel block from a reference may be constrained based on the reference. In as aspect, if the reference is a temporally interpolated virtual reference frame with corresponding time close to the time of the current pixel block, the motion vector for prediction may be constrained magnitude and/or precision. In another aspect, a bitstream syntax for encoding the constrained motion vector may also be constrained. In this manner, the techniques proposed herein contribute to improved coding efficiencies.