InterDigital CE Patent Holdings

In various implementations, methods and devices are disclosed that process one or more syntax elements for reconstructing at least one part of an picture end-to-end coded using a neural network. The one or more syntax elements comprise at least one of an indication (ph_nn_allowed_flag) enabling end-to-end neural network coding for at least a part of the picture; an indication specifying (ph_only_nn_slice) that all parts of the picture are end-to-end neural network coded; or an indication (sh_nn_flag) enabling end-to-end neural network coding at a given level of the picture, the given level comprises a slice level, a coding tree level, a block level.

Methods and apparatus are provided for adaptive quantization parameter selection based on a cost using a neural network. In one embodiment, the neural network can be trained from a training database comprising numerous coding units or blocks. In another embodiment, a rate distortion cost is determined for combinations of quantization parameter and split configuration. The quantization parameter with a minimal rate distortion cost for a particular split is used for encoding. In another embodiment, rate and distortion are separately approximated, resulting in lowest rate cost and lowest distortion cost for each possible type of split.

Method and apparatus are provided wherein one or more syntax elements are decoded providing for reconstructing at least one part of an image or a 3D object using at least one hybrid Implicit Neural Representation network, the at least one hybrid Implicit Neural Representation network comprising at least one mapping transformation that maps input coordinates of the image or 3D object to at least two successive grids of features, and an INR network that takes as input, features of the at least one grid of features and outputs values of the image or 3D object at the input coordinates, wherein dimensions of a current grid of the at least two successive grids of features are derived from a first grid of the at least two successive grids of features based on the one or more syntax elements. An extension of at least one dimension of a first grid of the at least two successive grids of features is determined providing an extended dimension of the first grid of the at least two successive grids of features such that dimensions of grids of the at least two successive grids of features are deteremined using the extended dimension of the first grid and input coordinates are mapped to grid positions of the grids of the at least two successive grids of features to reconstruct the at least one part of the image or 3D object.

Some embodiments of a method may include: obtaining avatar format data associated with an avatar; parsing the avatar format data for a textures structure; and performing a loop for each texture element of the textures structure, wherein the loop includes: determining a U-Dimensional (UDIM)-based extension exists within the textures structure; parsing data associated with the UDIM-based extension; determining a u-coordinate (uValue) attribute exists within the texture element; parsing the uValue attribute; determining a v-coordinate (vValue) attribute exists within the texture element; parsing the vValue attribute; determining a source attribute exists within the texture element; and parsing the source attribute.

Method and apparatus for encoding and decoding an input signal, for example an image, a video or a 3D object, using a hybrid Implicit Neural Representation network are provided. The method comprises obtaining one or more syntax elements providing for reconstructing at least one part of the input signal using at least one hybrid Implicit Neural Representation network, the at least one hybrid Implicit Neural Representation network comprising: at least one mapping transformation that maps input coordinates of the image or 3D object to at least one grid of features, and an INR network that takes as input, features of the at least one grid of features and outputs values of the image or 3D object at the input coordinates, wherein the one or more syntax elements comprises a first syntax element indicating whether a first grid of features of the at least one grid of features has same dimensions as dimensions of the input coordinates, and signaling the one or more syntax elements in a bitstream.

A method comprising:obtaining (40) at least two video sequences;determining (41) a first reference level value for each video sequence, each first reference level value being representative of a luminance level of a same area of interest in each video sequence;obtaining (43) a second reference level value; and,modifying (44) at least one of the video sequences so that, for each modified video sequence, the luminance level of the same area of interest in the video sequence is equal to the second reference level value using a conversion curve depending on the first reference level value of the video sequence and on the second reference level value.

In example methods and apparatus, during an extended reality, XR, spatial computing session between the UE and a server, at least a first sensor data stream is provided from the UE to the server using at least a first sensor data setting. Information indicating an updated first sensor data setting is communicated between the UE and the server. In response to the communication between the UE and the server, the first sensor data setting is updated to the updated first sensor data setting. The UE then continues to provide the first sensor data stream from the UE to the server using the updated first sensor data setting.

Methods and devices enable an adaptive haptic experience that endows users with adjustments to their profile or to their physiological reactions. In such context, data representative of a haptic experience not only contains haptic data but also affective data. This additional data indicates the desired affective response of the user during the haptic experience. Two embodiments for adapting a haptic signal to provide consistency in users perceived affect to a haptic stimulus are proposed herein. In a first embodiment, the adaptation is based on an affective haptic profile that describes the user's preferred settings regarding haptic content and affect and comprises parameters allowing to adapt a haptic signal accordingly. In a second embodiment, the adaptation takes into account real time data extracted from users' physiological reactions, thus delivering a physiological closed loop control system.

In one general aspect, a method is provided that includes receiving from a first user device, an animation data stream for animating an avatar associated with the first user device, the avatar being part of a shared scene of the multi-user avatar-based session between the first user device and at least one second user device; determining for the at least one second user device, whether the at least one second user device has capabilities to animate the avatar associated with the first user device using the animation data stream. If the at least one second user device has capabilities to animate the avatar, the animation data stream is forwarded to the at least one second user device, otherwise the animation data stream is converted based on at least one second user device's capability before being sent to the at least one second user device.

In one implementation, a method of decoding a signal comprises: decoding information indicating partitioning of the signal into a plurality of partitions; for each partition of the plurality of partitions: decoding information indicating association of the partition to a respective processing model of a plurality of processing models; decoding latent variables; processing the latent variables based on the respective processing models to form processed latent variables; and reconstructing the signal from the processed latent variables. At the encoder side, a method comprises: encoding information indicating partitioning of the signal into a plurality of partitions; for each partition of the plurality of partitions: encoding information indicating association of the partition to a respective processing model of a plurality of processing models; obtaining latent variables associated with coordinates of samples of the signal; encoding the latent variables; and encoding model parameters for the plurality of processing models.