Abstract: A method including, in a training phase, training a gaze prediction model including a first model and a second model, the first model and the second model being configured in conjunction to predict segmentation data based on training data, training a third model together with the first model and the second model, the third model being configured to predict a training characteristic using an output of the first model based on the training data, and in an operational phase, receiving operational data and predicting an operational characteristic using the trained first model and the trained third model.

Abstract: The present invention, with respect to a method of processing video data, provides a method of processing video data, provides a method characterized by comprising the steps of: performing a clustering for the video data; generating at least one data cluster as a result of the clustering; generating at least one Graph laplacian matrix corresponding to the at least one data cluster; performing conversion optimization on the basis of multiple graph-based models, wherein the multiple graph-based models respectively include at least one graph laplacian matrix; and generating an optimized conversion matrix according to the results of performing the conversion optimization.

Abstract: The present invention, with respect to a method of processing video data, provides a method of processing video data, provides a method characterized by comprising the steps of: performing a clustering for the video data; generating at least one data cluster as a result of the clustering; generating at least one Graph laplacian matrix corresponding to the at least one data cluster; performing conversion optimization on the basis of multiple graph-based models, wherein the multiple graph-based models respectively include at least one graph laplacian matrix; and generating an optimized conversion matrix according to the results of performing the conversion optimization.

Abstract: A handheld user device includes a monocular camera to capture a feed of images of a local scene and a processor to select, from the feed, a keyframe and perform, for a first image from the feed, stereo matching using the first image, the keyframe, and a relative pose based on a pose associated with the first image and a pose associated with the keyframe to generate a sparse disparity map representing disparities between the first image and the keyframe. The processor further is to determine a dense depth map from the disparity map using a bilateral solver algorithm, and process a viewfinder image generated from a second image of the feed with occlusion rendering based on the depth map to incorporate one or more virtual objects into the viewfinder image to generate an AR viewfinder image. Further, the processor is to provide the AR viewfinder image for display.

Abstract: The present invention provides a method for performing a transform, the method comprising the steps of: deriving a row transform set, a column transform set, and a permutation matrix on the basis of a given transform matrix (H) and error tolerance parameter; obtaining a row-column transform (RCT) coefficient on the basis of the row transform set, the column transform set, and the permutation matrix; and performing a quantization and an entropy encoding on the RCT coefficient, wherein the permutation matrix represents a matrix obtained by permutating a row of an identity matrix.

Abstract: Disclosed is a method allowing enhanced prediction of a video signal, the method comprising the steps of: entropy-decoding a neighboring block adjacent to a target block; inverse-quantizing the entropy-decoded neighboring block; acquiring the modified neighboring block by carrying out a modified inverse transform on an inverse-quantized transform coefficient vector of the neighboring block; and generating a prediction block for the target block based on the modified neighboring block, wherein the modified inverse transform applies a different scaling matrix for each pixel location reconstructed for the inverse-quantized transform coefficient vector of the neighboring block.

Abstract: Provided is a method for processing a video signal on the basis of an arbitrary partition transform, the method comprising the steps of: determining a partition set for input data, the partition set indicating a set of partitions divided from a data block on the basis of a scanning pattern; determining a transform matrix set corresponding to the partition set; obtaining an arbitrary partition transform vector for each partition by applying the transform matrix set to the partition set; and obtaining the arbitrary partition transform coefficient for the input data by inverse-mapping the arbitrary partition transform vector.

Abstract: A handheld user device includes a monocular camera to capture a feed of images of a local scene and a processor to select, from the feed, a keyframe and perform, for a first image from the feed, stereo matching using the first image, the keyframe, and a relative pose based on a pose associated with the first image and a pose associated with the keyframe to generate a sparse disparity map representing disparities between the first image and the keyframe. The processor further is to determine a dense depth map from the disparity map using a bilateral solver algorithm, and process a viewfinder image generated from a second image of the feed with occlusion rendering based on the depth map to incorporate one or more virtual objects into the viewfinder image to generate an AR viewfinder image. Further, the processor is to provide the AR viewfinder image for display.

Abstract: The present invention provides a method for encoding a video signal by using an enhanced prediction signal, the method comprising the steps of: generating a prediction signal for a current block, wherein the prediction signal is generated by applying a prediction weight to a reconstructed pixel of the current block and a pixel of a motion-compensated previously decoded block; obtaining an optimal quantized transform coefficient for the current block through rate-distortion optimization, on the basis of the prediction signal; and performing entropy encoding of the optimal quantized transform coefficient.

Abstract: The present invention, with respect to a method of processing video data, provides a method of processing video data, provides a method characterized by comprising the steps of: performing a clustering for the video data; generating at least one data cluster as a result of the clustering; generating at least one Graph laplacian matrix corresponding to the at least one data cluster; performing conversion optimization on the basis of multiple graph-based models, wherein the multiple graph-based models respectively include at least one graph laplacian matrix; and generating an optimized conversion matrix according to the results of performing the conversion optimization.

Abstract: The present invention provides a method for performing transformation using a layered Givens transform (LGT), comprising the steps of: deriving at least one rotation layer and at least one permutation layer on the basis of a given transform matrix (H) and a given error parameter; acquiring a LGT coefficient on the basis of the rotation layer and the permutation layer; and quantizing and entropy-encoding the LGT coefficient, wherein the permutation layer comprises a permutation matrix obtained by permuting a row of an identity matrix.

Abstract: The present invention provides a method for encoding a video signal by using a single optimized graph, comprising the steps of: obtaining a residual block; generating graphs from the residual block; generating an optimized graph and an optimized transform by combining the graphs, wherein the graphs are combined on the basis of an optimization step; and performing a transform for the residual block on the basis of the optimized graph and the optimized transform.

Abstract: A method for encoding a video signal includes: a step of selecting a transform parameter set via an optimization process for a target block, wherein the transform parameter set comprises a vertical transform set, a horizontal transform set and a transform order parameter set; a step of updating, on the basis of the selected transform parameter set, a transform block having a singleton coefficient, wherein the single ton coefficient represents a single coefficient obtained on the basis of the transform order parameter set; a step of obtaining a transform coefficient for the target block on the basis of the updated transform block; a step of quantizing the transform coefficient; and a step of entropy encoding the quantized transform coefficient.

Abstract: The present invention provides a method for performing transformation using a layered Givens transform (LGT), comprising the steps of: deriving at least one rotation layer and at least one permutation layer on the basis of a given transform matrix (H) and a given error parameter; acquiring a LGT coefficient on the basis of the rotation layer and the permutation layer; and quantizing and entropy-encoding the LGT coefficient, wherein the permutation layer comprises a permutation matrix obtained by permuting a row of an identity matrix.

Abstract: Disclosed is a method allowing enhanced prediction of a video signal, the method comprising the steps of: entropy-decoding a neighboring block adjacent to a target block; inverse-quantizing the entropy-decoded neighboring block; acquiring the modified neighboring block by carrying out a modified inverse transform on an inverse-quantized transform coefficient vector of the neighboring block; and generating a prediction block for the target block based on the modified neighboring block, wherein the modified inverse transform applies a different scaling matrix for each pixel location reconstructed for the inverse-quantized transform coefficient vector of the neighboring block.

Abstract: Provided is a method for processing a video signal on the basis of an arbitrary partition transform, the method comprising the steps of: determining a partition set for input data, the partition set indicating a set of partitions divided from a data block on the basis of a scanning pattern; determining a transform matrix set corresponding to the partition set; obtaining an arbitrary partition transform vector for each partition by applying the transform matrix set to the partition set; and obtaining the arbitrary partition transform coefficient for the input data by inverse-mapping the arbitrary partition transform vector.

Abstract: The present invention provides a method for encoding a video signal by using an enhanced prediction signal, the method comprising the steps of: generating a prediction signal for a current block, wherein the prediction signal is generated by applying a prediction weight to a reconstructed pixel of the current block and a pixel of a motion-compensated previously decoded block; obtaining an optimal quantized transform coefficient for the current block through rate-distortion optimization, on the basis of the prediction signal; and performing entropy encoding of the optimal quantized transform coefficient.

Abstract: The present invention provides a method for encoding a video signal by using a single optimized graph, comprising the steps of: obtaining a residual block; generating graphs from the residual block; generating an optimized graph and an optimized transform by combining the graphs, wherein the graphs are combined on the basis of an optimization step; and performing a transform for the residual block on the basis of the optimized graph and the optimized transform.

Abstract: A method for performing enhanced prediction on video signals includes the steps of: performing entropy decoding and dequantization on a neighboring block of a target block; performing a scaling on a dequantized transform coefficient for the neighboring block by using a scaling matrix; acquiring a modified neighboring block by performing inverse transform on the scaled transform coefficient; and generating a prediction block for the target block on the basis of the modified neighboring block.

Abstract: Disclosed herein is a method of calculating a displacement vector of a target region; determining an anchor region by using the calculated displacement vector; predicting a target region by linearly filtering the anchor region using a designed filter; and generating a prediction error by using the predicted target region.