Abstract: This disclosure relates generally to image processing and specifically to processing panorama images using neural networks to generate depth maps, layouts, semantic maps or the like with reduced distortion and improved continuity. Methods and systems are described for generating such maps by leveraging several essential properties of these panorama images and by using a panorama panel representation and a neural network framework. A panel geometry embedding network is incorporated for encoding both the local and global geometric features of the panels in order to reduce negative impact of panoramic distortion. A local-to-global transformer network is also incorporated for capturing geometric context and aggregating local information within a panel and panel-wise global context.

Abstract: In a method of encoding performed in an encoder, a block is partitioned into a first part and a second part based on a geometry partition mode (GPM). Inter prediction is performed on the first part and the second part of the block based on merge mode with motion vector difference (MMVD). Whether a first motion vector (MV) for the first part is identical to a second MV for the second part is determined based on a first distance index and a first direction index and a second distance index and a second direction index. The block is encoded in a bitstream based on the first MV and the second MV when the first MV is not identical to the second MV, and based on the first MV and a third MV for the second part of the block when the first MV is identical to the second MV.

Abstract: Method, apparatus, and non-transitory storage medium for hybrid acoustic howling suppression based on a frequency filter model and a deep neural network are provided. The method may include receiving a speech signal, the speech signal including target speech, feedback, and noise, and inputting the speech signal into a trained hybrid neural-network based howling suppression model, wherein the trained hybrid neural-network based howling suppression model is trained using training speech signal and pre-processed acoustic feedback from a first frequency filter model. The method may also include generating an enhanced speech signal with suppressed howling as an output of the trained hybrid neural-network based howling suppression model, wherein the enhanced speech signal is used to update parameters of the first frequency filter model.

Abstract: In a method, a first area and a second area of a current frame is determined when the current frame is a GDR frame. The first area is independently coded and the second area is coded with dependency. When a current block in the current frame is coded by an intraTMP mode or an IBC mode, a search range of the intraTMP mode or the IBC mode is determined such that the search range is only in the first area of the current frame. A reference block is determined from a plurality of candidate reference blocks in the search range. The current block is encoded in a bitstream based on the determined reference block.

Abstract: A method includes receiving, by a 5th generation media streaming (5GMS) client for an uplink streaming session, a media entry point and one or more operation point parameters; determining, by the 5GMS client, a plurality of available service descriptions for the media entry point; transmitting, from the 5GMS client to a 5GMS application, the plurality of available service descriptions; receiving, by the 5GMS client from the 5GMS application, a selected service description from the plurality of available service descriptions; selecting, by the 5GMS client, a dynamic policy based on a plurality of Service Operation Point parameters associated with the selected service description; transmitting, by the 5GMS client to a 5GMS application function (AF), the dynamic policy and the plurality of Service Operation Point parameters associated with the selected service description; establishing, with the 5GMS AF, the uplink streaming session.

Abstract: Method, apparatus, and non-transitory storage medium for training a deep neural-network model jointly for acoustic echo suppression and acoustic howling suppression are provided. The method may include generating a teacher speech signal for training the deep neural-network model based on a input speech from a speech system and at least one reference signal. The deep neural-network model is trained jointly for both acoustic echo suppression and acoustic howling suppression by using the teacher speech signal and a correlation loss. During training of the deep neural-network model, the training task formulates a recurrent feedback suppression process as an instantaneous speech separation task using the teacher-forced training strategy.

Abstract: A method and apparatus comprising computer code configured to cause a processor or processors to envelope a message, of one or more federated learning messages, by a control message format, the control message format comprising a plurality of fields respectively indicating ones of an identifier of the message, a size of the message, a type of the message, and a body of the message, and control the artificial intelligence/machine learning federated learning based on the message enveloped by the control message format.

Abstract: In a method of video processing in a decoder, information of a coding block in a current picture is decoded from a bitstream. The information indicates a bi-prediction mode. A first motion vector associated with a first reference picture and a second motion vector associated with a second reference picture for a bi-prediction of the coding block are determined. A first reference template in the first reference picture is determined based on a current template of the coding block and the first motion vector. A second reference template in the second reference picture is determined based on the current template of the coding block and the second motion vector. A weight to be applied in the bi-prediction mode is calculated based on the current template and a difference between the first reference template and the second reference template. The coding block is reconstructed using the bi-prediction with the calculated weight.

Abstract: A video decoding method includes: obtaining a bitstream including a plurality of coded frames of a video signal; decoding each of the plurality of coded frames into a plurality of super blocks and each of the plurality of super blocks into a plurality of residual blocks; recovering a coded block (CB) for each of the plurality of residual blocks based on multiple reference line intra prediction (MRLP) flags and reference samples included in each coded frame, wherein multiple reference lines are divided into above-side reference lines and left-side reference lines and one above-side reference line and one left-side reference line are selected for intra prediction; reconstructing each frame of the video signal by storing the recovered CB for each of the plurality of residual blocks in a frame buffer; and continuously outputting the reconstructed frames to restore the video signal.

Abstract: A method of processing media content in Moving Picture Experts Group (MPEG) Network Based Media Processing (NBMP) may include obtaining, from an NBMP source, a workflow having a workflow descriptor (WD) indicating a workflow descriptor document (WDD); based on the workflow, obtaining a task having a task descriptor (TD) indicating a task descriptor document (TDD); based on the task, obtaining, from a function repository, a function having a function descriptor (FD) indicating a function descriptor document (FDD); and processing the media content, using the workflow, the task, and the function.

Abstract: The various embodiments described herein include methods and systems for coding video. In one aspect, a method includes obtaining encoded video data comprising a plurality of blocks and obtaining a motion vector predictor (MVP) candidate block from a MVP list based on a MVP index. The method further includes in accordance with a determination that a block of the plurality of blocks is designated for a warp extend mode, determining whether the MVP candidate block is suitable for the warp extend mode. The method also includes, in accordance with a determination that the MVP candidate block is not suitable for the warp extend mode, identifying a backup MVP candidate block that is suitable for the warp extend mode. The method further includes obtaining a warp model from the backup MVP candidate block; and performing a warp extend operation on the block using the warp model.

Abstract: An apparatus for point cloud coding, includes processing circuitry that receives a coded bitstream for a point cloud. The coded bitstream includes encoded data for nodes in an octree structure for the point cloud corresponding to three dimensional (3D) partitions of a space of the point cloud, node sizes of the nodes being associated with sizes of the corresponding 3D partitions of the nodes. The processing circuitry decodes, from the coded bitstream, a first set of occupancy codes for a first set of nodes in the nodes using a first coding order and a second set of occupancy codes for a second set of nodes in the nodes using a second coding order that is different from the first coding order. Further, the processing circuitry reconstructs the octree structure based on at least the first set of occupancy codes and the second set of occupancy codes.

Abstract: A Method of decoding an encoded video bitstream using at least one processor includes: obtaining an encoded video bitstream, the encoded video bitstream including encoded color components; entropy parsing the encoded color components; dequantizing the color components and obtaining transform coefficients of the color components; applying a joint components secondary transform (JCST) on the transform coefficients of the color components, thereby generating JCST outputs; performing a backward transform on the JCST outputs, thereby obtaining residual components of the color components; and decoding the encoded video bitstream based on the residual components of the color components.

Abstract: Aspects of the disclosure provide methods and apparatuses for video encoding/decoding. An apparatus for video decoding includes processing circuitry that decodes prediction information for a current block in a current picture that is included in a coded video sequence. The prediction information indicates an intra prediction mode of the current block. The processing circuitry determines a position dependent prediction combination (PDPC) process according to the intra prediction mode of the current block indicated by the prediction information. Further, the processing circuitry reconstructs the current block based on the determined PDPC process. A same PDPC process is applied to intra prediction modes adjacent to diagonal intra prediction modes.

Abstract: A method including receiving an input comprising natural language texts; segmenting the natural language texts into sections; summarizing the natural language texts; developing a first model based on the plurality of sections and the summary of the natural language texts; identifying one or more salient sentences within the natural language texts using the first model; determining a sentence quality score based on how informative a salient sentence is; determining a sentence similarity score based on a salient sentence's similarity to another salient sentence; developing a second model based on the sentence quality score and the sentence similarity score; combining the first model and the second model into a final model; selecting sentences based on the final model; and generating an extractive summarization using the selected sentences.

Abstract: This disclosure relates generally to video coding/decoding and particularly for providing extension to block adaptive weighted prediction (BAWP). One method includes receiving a coded video bitstream; determining, based on a syntax element signaled in the coded video bitstream, a prediction mode for predicting the current block based on a reference block, wherein the prediction mode comprises a linear equation; deriving a scaling factor for the current block, from at least one of the following: multiple scaling factors of neighboring blocks with respect to the current block, or a stored scaling factor bank; and reconstructing the current block based on the reference block and the identified scaling factor according to the linear equation.

Abstract: This disclosure relates generally to video coding/decoding and particularly for providing extension to block adaptive weighted prediction (BAWP) with multiple motion vectors. One method includes receiving a coded video bitstream; identifying, from the coded video bitstream, a first motion vector corresponding to a first reference block and a second motion vector corresponding to a second reference block; obtaining a first scaling factor corresponding to the first motion vector and a second scaling factor corresponding to the second motion vector by parsing the coded video bitstream; generating a first predicted block based on the first scaling factor and the first reference block according to a first linear equation; generating a second predicted block based on the second reference block according to a second linear equation; and reconstructing the current block based on the first predicted block and the second predicted block.

Abstract: A method and apparatus for media decoding by a decoder include decoding a first indication indicative of a first conformance point of a coded video sequence. A second indication indicative of a second conformance point of the coded video sequence is decoded. It is determined whether the coded video sequence is decodable by the decoder based on at least one of the first indication and the second indication. The coded video sequence is selectively decoded based on determining whether the decoded video sequence is decodable by the decoder.

Abstract: An affine motion estimation (ME) is performed on a current block to determine affine parameters of the current block in a pre-analysis stage. The affine ME is performed based on at least one of (i) a down-scaled current picture and one or more down-scaled reference pictures of the down-scaled current picture, (ii) a simplified configuration, or (iii) a fixed block size of the current block being equal to or larger than a threshold. The affine parameters of the current block determined in the pre-analysis stage are stored. Affine parameters of the current block are determined based on the affine parameters stored in the pre-analysis stage. The current block is reconstructed based on the determined affine parameters of the current block.

Abstract: Aspects of the disclosure provide methods and apparatuses for video encoding/decoding. In some examples, an apparatus for video encoding includes processing circuitry. The processing circuitry determines whether a current block in a current picture is a small block based on a block size threshold. The processing circuitry constructs a motion vector predictor list for the current block based on whether the current block is the small block, at least one redundancy check with a motion vector candidate in the motion vector predictor list being performed in the construction of the motion vector predictor list based on whether the current block is the small block. The processing circuitry encodes the current block based on the constructed motion vector predictor list.