Abstract: In an example method, a system receives a plurality of frames of a video, and generates a data structure representing the video and representing a plurality of temporal layers. Generating the data structure includes: (i) determining a plurality of quality levels for presenting the video, where each of the quality levels corresponds to a different respective sampling period for sampling the frames of the video, (ii) assigning, based on the sampling periods, each of the frames to a respective one of the temporal layers of the data structure, and (iii) indicating, in the data structure, one or more relationships between (a) at least one the frames assigned to at least one of the temporal layers of the data structure, and (b) at least another one of the frames assigned to at least another one of the temporal layers of the data structure. Further, the system outputs the data structure.

Abstract: A system comprises an encoder configured to compress attribute information and/or spatial for a point cloud and/or a decoder configured to decompress compressed attribute and/or spatial information for the point cloud. To compress the attribute and/or spatial information, the encoder is configured to convert a point cloud into an image based representation. Also, the decoder is configured to generate a decompressed point cloud based on an image based representation of a point cloud. A processing/filtering element utilizes occupancy map information and/or auxiliary patch information to determine relationships between patches in image frames and adjusts encoding/decoding and/or filtering or pre/post-processing parameters based on the determined relationships.

Abstract: In an example method, a system accesses first input data and a machine learning architecture. The machine learning architecture includes a first module having a first neural network, a second module having a second neural network, and a third module having a third neural network. The system generates a first feature set representing a first portion of the first input data using the first neural network, and a second feature set representing a second portion of the first input data using the second neural network. The system generates, using the third neural network, first output data based on the first feature set and the second feature set.

Abstract: A filtering system for video coders and decoders is disclosed that includes a feature detector having an input for samples reconstructed from coded video data representing a color component of source video, and having an output for data identifying a feature recognized therefrom, an offset calculator having an input for the feature identification data from the feature detector and having an output for a filter offset, and a filter having an input for the filter offset from the offset calculator and an input for the reconstructed samples, and having an output for filtered samples. The filtering system is expected to improve operations of video coder/decoder filtering systems by selecting filtering offsets from analysis of recovered video data in a common color plane as the samples that will be filtered.

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: 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: In communication applications, aggregate source image data at a transmitter exceeds the data that is needed to display a rendering of a viewport at a receiver. Improved streaming techniques that include estimating a location of a viewport at a future time. According to such techniques, the viewport may represent a portion of an image from a multi-directional video to be displayed at the future time, and tile(s) of the image may be identified in which the viewport is estimated to be located. In these techniques, the image data of tile(s) in which the viewport is estimated to be located may be requested at a first service tier, and the other tile in which the viewport is not estimated to be located may be requested at a second service tier, lower than the first service tier.

Abstract: Improved neural-network-based image and video coding techniques are presented, including hybrid techniques that include both tools of a host codec and neural-network-based tools. In these improved techniques, the host coding tools may include conventional video coding standards such H.266 (VVC). In an aspects, source frames may be partitioned and either host or neural-network-based tools may be selected per partition. Coding parameter decisions for a partition may be constrained based on the partitioning and coding tool selection. Rate control for host and neural network tools may be combined. Multi-stage processing of neural network output may use a checkerboard prediction pattern.

Abstract: In an example method, a system receives a plurality of frames of a video, and generates a data structure representing the video and representing a plurality of temporal layers. Generating the data structure includes: (i) determining a plurality of quality levels for presenting the video, where each of the quality levels corresponds to a different respective sampling period for sampling the frames of the video, (ii) assigning, based on the sampling periods, each of the frames to a respective one of the temporal layers of the data structure, and (iii) indicating, in the data structure, one or more relationships between (a) at least one the frames assigned to at least one of the temporal layers of the data structure, and (b) at least another one of the frames assigned to at least another one of the temporal layers of the data structure. Further, the system outputs the data structure.

Abstract: Techniques are disclosed for coding video data in which frames from a video source are partitioned into a plurality of tiles of common size, and the tiles are coded as a virtual video sequence according to motion-compensated prediction, each tile treated as having respective temporal location of the virtual video sequence. The coding scheme permits relative allocation of coding resources to tiles that are likely to have greater significance in a video coding session, which may lead to certain tiles that have low complexity or low motion content to be skipped during coding of the tiles for select source frames. Moreover, coding of the tiles may be ordered to achieve low coding latencies during a coding session.

Abstract: Support for additional components may be specified in a coding scheme for image data. A layer of a coding scheme that specifies color components may also specify additional components. Characteristics of the components may be specified in the same layer or a different layer of the coding scheme. An encoder or decoder may identify the specified components and determine the respective characteristics to perform encoding and decoding of image data.

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 coding and decoding video data using object recognition and object modeling as a basis of coding and error recovery. A video decoder may decode coded video data received from a channel. The video decoder may perform object recognition on decoded video data obtained therefrom, and, when an object is recognized in the decoded video data, the video decoder may generate a model representing the recognized object. It may store data representing the model locally. The video decoder may communicate the model data to an encoder, which may form a basis of error mitigation and recovery. The video decoder also may monitor deviation patterns in the object model and associated patterns in audio content; if/when video decoding is suspended due to operational errors, the video decoder may generate simulated video data by analyzing audio data received during the suspension period and developing video data from the data model and deviation(s) associated with patterns detected from the audio data.

Abstract: The present disclosure describes techniques for coding and decoding video in which a plurality of coding hypotheses are developed for an input pixel block of frame content. Each coding hypothesis may include generation of prediction data for the input pixel block according to a respective prediction search. The input pixel block may be coded with reference to a prediction block formed from prediction data derived according to plurality of hypotheses. Data of the coded pixel block may be transmitted to a decoder along with data identifying a number of the hypotheses used during the coding to a channel. At a decoder, an inverse process may be performed, which may include generation of a counterpart prediction block from prediction data derived according to the hypothesis identified with the coded pixel block data, then decoding of the coded pixel block according to the prediction data.

Abstract: Video compression and decompression techniques are disclosed that provide improved bandwidth control for video compression and decompression systems. In particular, video coding and decoding techniques quantize input video in multiple dimensions. According to these techniques, pixel residuals may be generated from a comparison of an array of input data to an array of prediction data. The pixel residuals may be quantized in a first dimension. After the quantization, the quantized pixel residuals may be transformed to an array of transform coefficients. The transform coefficients may be quantized in a second dimension and entropy coded. Decoding techniques invert these processes. In still other embodiments, multiple quantizers may be provided upstream of the transform stage, either in parallel or in cascade, which provide greater flexibility to video coders to quantize data in different dimensions in an effort to balance the competing interest in compression efficiency and quality of reconstructed video.

Abstract: A system comprises an encoder configured to compress attribute information and/or spatial for a point cloud and/or a decoder configured to decompress compressed attribute and/or spatial information for the point cloud. To compress the attribute and/or spatial information, the encoder is configured to convert a point cloud into an image based representation. Also, the decoder is configured to generate a decompressed point cloud based on an image based representation of a point cloud. A closed-loop color conversion process is used to improve compression while taking into consideration distortion introduced throughout the point cloud compression process.

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: An encoder or decoder can perform enhanced motion vector prediction by receiving an input block of data for encoding or decoding and accessing stored motion information for at least one other block of data. Based on the stored motion information, the encoder or decoder can generate a list of one or more motion vector predictor candidates for the input block in accordance with an adaptive list construction order. The encoder or decoder can predict a motion vector for the input block based on at least one of the one or more motion vector predictor candidates.

Abstract: Embodiments of the present disclosure provide systems and methods for perspective shifting in a video conferencing session. In one exemplary method, a video stream may be generated. A foreground element may be identified in a frame of the video stream and distinguished from a background element of the frame. Data may be received representing a viewing condition at a terminal that will display the generated video stream. The frame of the video stream may be modified based on the received data to shift of the foreground element relative to the background element. The modified video stream may be displayed at the displaying terminal.

Abstract: The present disclosure describes techniques for coding and decoding video in which a plurality of coding hypotheses are developed for an input pixel block of frame content. Each coding hypothesis may include generation of prediction data for the input pixel block according to a respective prediction search. The input pixel block may be coded with reference to a prediction block formed from prediction data derived according to plurality of hypotheses. Data of the coded pixel block may be transmitted to a decoder along with data identifying a number of the hypotheses used during the coding to a channel. At a decoder, an inverse process may be performed, which may include generation of a counterpart prediction block from prediction data derived according to the hypothesis identified with the coded pixel block data, then decoding of the coded pixel block according to the prediction data.