Abstract: A label switching router for a switching network configured for source routing can be configured to modify one or more reported properties of a link associated with that router based on one or more link quality metrics. For example, a router can include a link with available bandwidth of 10 Gbps, a value that can be used by a headend when evaluating nodes of the network against a set of constraints required by a particular requested connection or tunnel. A link of that router may exhibit an increased bit error rate which can be used by the label switching router to artificially deflate available bandwidth, thereby reducing the number of label switching paths including the router are selected by the headend.

Abstract: Techniques for synchronized out-of-order live video encoding are described. As one example, a computer-implemented method includes receiving a first real-time encoded live video, that is in a coded order that is different than a display order, at a content delivery service from a contribution encoder at a different location than the content delivery service, performing, in the coded order, a real-time decode of the first real-time encoded live video by a distribution decoder of the content delivery service to generate a decoded video in the coded order, performing, in the coded order, a real-time encode of the decoded video by a distribution encoder of the content delivery service to generate a second real-time encoded video in the coded order without waiting on a reordering of the decoded video into the display order, and transmitting the second real-time encoded video in the coded order to a client device.

Abstract: A method for compressing display data is disclosed. The method comprises performing a wavelet transformation to obtain a general approximation coefficient and a plurality of detail coefficients for a group of pixels; determining whether to prioritise transmission of the general approximation coefficient over transmission of the detail coefficients based on whether there are sufficient resources available to enable a corresponding image frame to be ready for display and/or based on a time since the detail coefficients for a corresponding group of pixels were previously transmitted.

Abstract: A system for remote identification of users. The system uses deep learning techniques for authenticating a user from an identification document and using automated verification of identification documents. Identification documents may be authenticated by validating security features. The system may determine features expected in a valid identification document and determine whether those features are present, employing techniques, such as determining whether direction-sensitive features are present. Liveness of a user indicated by the identification document may be determined with a deep learning model trained for identification of facial spoofing attacks.

Abstract: A data-sparsity homogenizer includes a plurality of multiplexers and a controller. The plurality of multiplexers receives 2N bit streams of non-homogenous sparse data in which the non-homogenous sparse data includes non-zero value data clumped together. The plurality of multiplexers is arranged in 2N rows and N columns. Each input of a multiplexer in a first column receives a respective bit stream of the 2N bit streams of non-homogenized sparse data, and the multiplexers in a last column output 2N bit streams of sparse data that is more homogenous than the non-homogenous sparse data of the 2N bit streams. The controller controls the plurality of multiplexers so that the multiplexers in the last column output the 2N channels of bit streams of sparse data that is more homogeneous than the non-homogenous sparse data of the 2N bit streams.

Abstract: Provided are a display driving chip, a display apparatus and a display driving method. The display driving chip includes a gaze tracker, an image receiver, an image parser and an image driver, wherein the gaze tracker is configured to achieve real-time locating for the pupil center of human eyes and real-time calculation for gaze point coordinates, and output the gaze point coordinates of human eyes to the image parser; the image receiver is configured to receive image data to be played through a first image interface and input the image data to be played to the image parser; the image parser is configured to compress or decompress the image data to be played according to the gaze coordinates of human eyes and output the image data to the image driver.

Abstract: Apparatuses, methods, systems, and program products are disclosed for watermark security. An apparatus includes a watermark module configured to generate a digital watermark to be presented as part of a graphical interface based on data presented on the graphical interface. A digital watermark verifies an authenticity of data to be presented in a graphical interface. An apparatus includes a presentation module configured to embed the digital watermark into the graphical interface prior to the data being presented in the graphical interface such that the digital watermark is graphically imperceptible to a user, dynamically update the digital watermark during runtime in response to detecting a change in the at least a portion of the data that is encoded into the digital watermark, and re-embed the digital watermark into the graphical interface in response to the digital watermark being updated.

Abstract: Methods, systems, and non-transitory computer readable storage media are disclosed for utilizing deep learning to intelligently determine compression settings for compressing a digital image. For instance, the disclosed system utilizes a neural network to generate predicted perceptual quality values for compression settings on a compression quality scale. The disclosed system fits the predicted compression distortions to a perceptual distortion characteristic curve for interpolating predicted perceptual quality values across the compression settings on the compression quality scale. Additionally, the disclosed system then performs a search over the predicted perceptual quality values for the compression settings along the compression quality scale to select a compression setting based on a perceptual quality threshold. The disclosed system generates a compressed digital image according to compression parameters for the selected compression setting.

Abstract: A data-sparsity homogenizer includes a plurality of multiplexers and a controller. The plurality of multiplexers receives 2N bit streams of non-homogenous sparse data in which the non-homogenous sparse data includes non-zero value data clumped together. The plurality of multiplexers is arranged in 2N rows and N columns. Each input of a multiplexer in a first column receives a respective bit stream of the 2N bit streams of non-homogenized sparse data, and the multiplexers in a last column output 2N bit streams of sparse data that is more homogenous than the non-homogenous sparse data of the 2N bit streams. The controller controls the plurality of multiplexers so that the multiplexers in the last column output the 2N channels of bit streams of sparse data that is more homogeneous than the non-homogenous sparse data of the 2N bit streams.

Abstract: Systems and methods for encoding multiple video streams with digital watermarking for adaptive bitrate streaming in accordance with embodiments of the invention are disclosed.

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: Embodiments of an electronic device include an integrated circuit, a reconfigurable stream switch formed in the integrated circuit along with a plurality of convolution accelerators and a decompression unit coupled to the reconfigurable stream switch. The decompression unit decompresses encoded kernel data in real time during operation of convolutional neural network.

Abstract: In one example, an apparatus is provided. The apparatus includes a photodiode, a charge sensing unit, an analog-to-digital converter (ADC), and a controller. The controller is configured to: enable the photodiode to generate charge in response to incident light, accumulate at least a portion of the charge as residual charge until the photodiode becomes saturated by the residual charge, and transfer the remaining portion of the charge to the charge sensing unit as overflow charge if the photodiode becomes saturated by the residual charge. The controller is further configured to: generate, using the ADC, a first digital output based on the residual charge; after generating the first digital output, generate, using the ADC, a second digital output based on the overflow charge; and generate a digital representation of an intensity of the incident light based on at least one of the first digital output or the second digital output.

Abstract: If an image read by an image reading unit is formed by an image forming unit, a control unit synthesizes a first encoded image and a second encoded image with the image formed by the image forming unit to generate a synthetic image, the first encoded image representing encoded data where reading setting information used for the image reading unit to read the image is encoded, the second encoded image representing encoded data where forming setting information used for the image forming unit to form the image is encoded, the first encoded image being synthesized at a position where a direction in which the document is read by the image reading unit is specifiable, and the second encoded image being synthesized at a position where a direction in which the image is formed on the sheet is specifiable. The image forming unit forms the synthetic image on the sheet.

Abstract: The disclosure is related to a method and device for comparing media features, the method comprising: obtaining first media feature sequences of a first media object and second media feature sequences of a second media object, the first media feature sequence comprises a plurality of first media feature units arranged in sequence, and the second media feature sequence comprises a plurality of second media feature units arranged in sequence; determining unit similarities between the first media feature units and the second media feature units; determining a similarity matrix between the first media feature sequences and the second media feature sequences according to the unit similarities; determining a similarity of the first media object and the second media object according to the similarity matrix.

Abstract: An encoder includes circuitry and a memory coupled to the circuitry. The circuitry, in operation, determines whether or not a ternary split process of splitting a block into three sub blocks in a first direction parallel to a first longer side of the block is allowed by comparing a size of a second shorter side of the block to a minimum threshold value. The circuitry, responsive to the ternary split process being allowed, writes, into a bitstream, a split direction parameter indicative of a splitting direction. The circuitry, in operation, splits the block into a plurality of sub blocks in a direction indicated by the split direction parameter; and encodes the plurality of sub blocks.

Abstract: Deep feature extraction and training tools and processes may facilitate extraction and understanding of deep features utilized by deep learning models. For example, imaging data may be tessellated and masked to generate a plurality of masked images. The masked images may be processed by a deep learning model to generate a plurality of masked outputs. The masked outputs may be aggregated for each cell of the tessellated image and compared to an original output for the imaging data from the deep learning model. Individual cells and associated image regions having masked outputs that correspond to the original output may comprise deep features utilized by the deep learning model.

Abstract: In some examples, a method of decoding a point cloud includes decoding an initial QP value from an attribute parameter set. The method also includes determining a first QP value for a first component of an attribute of point cloud data from the initial QP value. The method further includes determining a QP offset value for a second component of the attribute of the point cloud data and determining a second QP value for the second component of the attribute from the first QP value and from the QP offset value. The method includes decoding the point cloud data based on the first QP value and further based on the second QP value.

Abstract: A fixed wireless access system is implemented using orthogonal time frequency space multiplexing (OTFS). Data transmissions to/from different devices share transmission resources using—delay Doppler multiplexing, time-frequency multiplexing, multiplexing at stream and/or layer level, and angular multiplexing. Time-frequency multiplexing is achieved by dividing the time-frequency plan into subgrids, with the subsampled time frequency grid being used to carry the OTFS data. Antenna implementations include a hemispherical antenna with multiple antenna elements arranged in an array to achieve multiplexing.

Abstract: Implementations set forth herein relate to an automated assistant that can control a camera according to one or more conditions specified by a user. A condition can be satisfied when, for example, the automated assistant detects a particular environment feature is apparent. In this way, the user can rely on the automated assistant to identify and capture certain moments without necessarily requiring the user to constantly monitor a viewing window of the camera. In some implementations, a condition for the automated assistant to capture media data can be based on application data and/or other contextual data that is associated with the automated assistant. For instance, a relationship between content in a camera viewing window and other content of an application interface can be a condition upon which the automated assistant captures certain media data using a camera.

Abstract: Methods and apparatus are provided for encoding and decoding binary sets using adaptive tree selection. In one exemplary encoding method embodiment, picture data is encoded for a block in a picture; in which one of a plurality of trees structures is selected to code a binary set of data for indicating coefficient significance for the block. In another exemplary encoding method embodiment, picture data is encoded for a block in a picture, in which one or more trees are used to encode a binary set of data for indicating coefficient significance for the block, the one or more trees each having a plurality of nodes, at least one of the nodes of the one or more trees being modified responsive to at least one parameter.

Abstract: A system comprises an encoder configured to compress attribute information for a point cloud and/or a decoder configured to decompress compressed attribute information for the point cloud. Attribute values for at least one starting point are included in a compressed attribute information file and attribute correction values used to correct predicted attribute values are included in the compressed attribute information file. Attribute values are predicted based, at least in part, on attribute values of neighboring points and distances between a particular point for whom an attribute value is being predicted and the neighboring points. The predicted attribute values are compared to attribute values of a point cloud prior to compression to determine attribute correction values. A decoder follows a similar prediction process as an encoder and corrects predicted values using attribute correction values included in a compressed attribute information file.

Abstract: A method for lossy image or video encoding, transmission and decoding, the method comprising the steps of: receiving an input image at a first computer system; encoding the first input training image using a first trained neural network to produce a latent representation; performing a quantization process on the latent representation to produce a quantized latent; entropy encoding the quantized latent using a probability distribution, wherein the probability distribution is defined using a tensor network; transmitting the entropy encoded quantized latent to a second computer system; entropy decoding the entropy encoded quantized latent using the probability distribution to retrieve the quantized latent; and decoding the quantized latent using a second trained neural network to produce an output image, wherein the output image is an approximation of the input training image.

Abstract: The present disclosure provides a computer-implemented method for encoding video. The method includes: generating training data based on one or more video sequences, the training data including a structure similarity index comprising at least one of structure similarity index (SSIM) or multi-scale-structural similarity index (MS-SSIM); training a rate-distortion optimization (RDO) model using the training data; processing the one or more video sequences using the rate-distortion optimization model.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for channel-wise autoregressive entropy models. In one aspect, a method includes processing data using a first encoder neural network to generate a latent representation of the data. The latent representation of data is processed by a quantizer and a second encoder neural network to generate a quantized latent representation of data and a latent representation of an entropy model. The latent representation of data is further processed into a plurality of slices of quantized latent representations of data wherein the slices are arranged in an ordinal sequence. A hyperprior processing network generates a hyperprior parameters and a compressed representation of the hyperprior parameters. For each slice, a corresponding compressed representation is generated using a corresponding slice processing network wherein a combination of the compressed representations form a compressed representation of the data.

Abstract: A method for lossy image or video encoding, transmission and decoding, the method comprising the steps of: receiving an input image at a first computer system; encoding the input image using a first trained neural network to produce a latent representation; identifying one or more regions of the input image associated with high visual sensitivity; encoding the one or more regions of the input image associated with high visual sensitivity using a second trained neural network to produce one or more region latent representations; performing a quantization process on the latent representation and the one or more region latent representations; transmitting the result of the quantization process to a second computer system; decoding the result of the quantization process to produce an output image, wherein the output image is an approximation of the input image.

Abstract: 2D machine readable symbologies are stylized and made aesthetically-appealing, facilitating their use to convey plural-symbol data on product packaging and other articles. In some arrangements, symbologies are stylized by geometric transformations (e.g., by multiple rotation and/or mirroring operations) to develop tiles having organized geometric structures. Such stylized symbologies can be decoded by existing code readers. A great variety of other features and arrangements are also detailed.

Abstract: Disclosed are techniques for processing satellite signals for computing a geospatial position. A plurality of GNSS signals are received from a plurality of GNSS satellites. An image is captured using an imaging device at least partially oriented toward the plurality of GNSS satellites. The image is segmented into a plurality of regions based on RF characteristics of objects in the image. An orientation of the image is determined. The plurality of GNSS satellites are projected onto the image based on the orientation of the image such that a corresponding region is identified for each of the plurality of GNSS satellites. Each of the plurality of GNSS signals is processed in accordance with the corresponding region.

Abstract: A method of processing image data for transmittal to a display device involves receiving a frame of image data, the frame being divided into tile groups composed of tiles of pixels, each having a number of colour component values of a first colour space. Each tile includes a number of colour component planes of the first colour space having the colour component values for the pixels forming the tile. Each tile group is processed in an execution unit, formed by arithmetic logic units (ALUs) and a local shared memory, where each ALU includes dedicated register space for use solely by the ALU, and each tile of each tile group is processed by a number of the ALUs of the execution unit.

Abstract: A filter for video coding is provided, where the filter is configured for processing a block for generation of a filtered block, and the block comprises a plurality of pixels. The filter includes one or more processor configured to: scan, according to a predefined scan template, to obtain a current pixel of the block and its neighboring pixels of the current pixel; obtain spectrum components by performing transform for the current pixel and its neighboring pixels; obtain filtered spectrum components based on a filtering parameter and the spectrum components; obtain filtered pixels by performing inverse transform for the filtered spectrum components; and generate a filtered block based on the filtered pixels. The filter is provided allowing improving the efficiency for video coding.

Abstract: Embodiments of the present disclosure provide a method and apparatus for training a lane line identifying model. The method includes: acquiring a first image of a lane line, the first image being generated using a generating model based on a second image of the lane line, the first image and the second image of the lane line being associated with different physical environments respectively; acquiring lane line information in the second image of the lane line; and training the lane line identifying model using the first image and the acquired lane line information of the lane line.

Abstract: Methods, devices, systems and computer software/program code products improve the reliability of scene reconstruction through the use of a persistent store or cache to retain scene information observed across one or more previous frames.

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. A point cloud attribute transfer algorithm may be used to determine distortion between an original point cloud and a reconstructed point cloud. Additionally, the point cloud attribute transfer algorithm may be used to select attribute values for a reconstructed point cloud such that distortion between an original point cloud and a reconstructed version of the original point cloud is minimized.

Abstract: Exemplary embodiments are directed to a system for selective replacement of an object in an image. The system includes an interface configured to receive as input an original image and a background image, and a processing device in communication with the interface. The processing device is configured to process the original image using a neural network to detect one or more objects in the original image, generate a neural network mask of the original image for the one or more objects in the original image, generate a filtered original image including the original image without the one or more objects, generate a modulated background image including a replacement background based on the neural network mask, and generate a combined image including the filtered original image combined with the modulated background image.

Abstract: An electronic apparatus includes a memory storing information on an artificial intelligence (AI) model comprising a plurality of layers, and a processor configured to obtain an output image that is processed from an input image using the AI model. The processor is configured to, based on a number of non-zero data values included in operation data output from a first layer among the plurality of layers, compress the operation data according to at least one of a plurality of coding modes and store the compressed data in an internal memory, obtain restoration data corresponding to the operation data by decompressing the compressed data stored in the internal memory, and provide the obtained restoration data to a second layer among the plurality of layers.

Abstract: The invention relates to a method and apparatus for image compression, particularly to an improved block-coding apparatus and method for image compression. Image compression systems such as JPEG and JPEG2000 are known and popular standards for image compression. Many of the advantageous features of JPEG2000 derive from the use of the EBCOT algorithm (Embedded Block-Coding with Optimized Truncation). One drawback of the JPEG2000 standards is computational complexity. This application discloses a relatively fast block-coding algorithm, particularly as compared with the standard JPEG2000 EBCOT algorithm. Computational complexity is reduced.

Abstract: A display device is proposed. The display device includes unit pixels in which red, white, blue, and green sub-pixels are sequentially arranged, and includes an image processor configured to detect text from an image signal input from outside, perform image processing on the text, and output the text, wherein in the unit pixels respectively corresponding to both edges of the text, the image processor may perform image processing so that predetermined sub-pixels adjacent to a central part of the text are driven in a light-emitting state.

Abstract: The present invention is directed to a method of transferring a chunk of data between unconnected computing devices. A first chunk of data present in a source computing device can be in a first format that is binary data or nonstandard characters. Encoding, by the source computing device, the first chunk of data to a second chunk of data. The second chunk of data is in a format that is of standard characters. Capturing an image of the second chunk of data by an optical sensor. Receiving of the image by the target computing device. Processing of the image, by the target computing device, to the second chunk of data. Decoding, by the target computing device, the secondary chunk of data to the first chunk of data.

Abstract: Disclosed is a system for dynamically classifying different data point sets within a point cloud with different classifications that may alter how data point sets with different classifications are processed, edited, and/or rendered. The system may generate a model based on a first set of relationships between a first set of data point elements that result in the first classification, and a second set of relationships between a second set of data point elements that result in the second classification. The system may compare the data point elements from unclassified data point sets against the first set of relationships and the second set of relationships in the model, and may assign the first classification to a particular unclassified data point set in response to the data point elements of the particular data point set having a threshold amount of the first set of relationships.

Abstract: Video decoding may include transform coefficient continuity smoothing, which may include coefficient continuity smoothing, defined correlation coefficient smoothing, pixel range projection, and luminance correlated chrominance resampling. Coefficient continuity smoothing may include obtaining encoded block data from the encoded bitstream, the encoded block data corresponding to a current block from the reconstructed frame, and generating reconstructed block data for the current block based on the encoded block data using transform coefficient continuity smoothing.

Abstract: A moving-image decoder includes a decoding means configured to decode a bitstream representing a moving image including a block encoded using intra-prediction, an inverse processing means configured to generate difference values in pixels of the block by inverse-quantizing and inverse-transforming a level value of the block obtained from a result of decoding, an intra-prediction means configured to generate predicted values of pixels of the block according to intra-prediction based on pixel values of reconstructed blocks around the block, and a reconstruction means configured to reconstruct the block on the basis of the generated difference values and the generated predicted values, wherein the intra-prediction means is configured to generate a reference line for use in the intra-prediction by synthesizing a plurality of lines of the reconstructed blocks around the block.

Abstract: In one example, a method for inserting a digital watermark in a signal includes obtaining the signal comprising a plurality of frames, inserting a first digital watermark in a first frame of the plurality of frames, inserting a second digital watermark in a second frame of the plurality of frames, wherein the second digital watermark differs from the first digital watermark in at least one way selected from a group of: a location within a respective frame, a number of bits, a pattern of bits, and a number of bits of a noise, and outputting a watermarked signal including the first digital watermark in the first frame and the second digital watermark in the second frame.

Abstract: A three-dimensional data encoding method includes: generating a bit sequence including N-bit information that is information of a current node included in an N-ary tree structure of three-dimensional points included in three-dimensional data and that indicates whether a three-dimensional point is present in each of child nodes belonging to the current node, where N is an integer greater than or equal to 2; generating (i) position information indicating a head position that is a position at which a predetermined code appears first in the bit sequence when the bit sequence is scanned in a predetermined scan order, and (ii) a remaining bit that is part of the bit sequence after the head position in the predetermined scan order; and encoding the position information and the remaining bit as information of the current node.

Abstract: In a data compression method of compressing data including at least one of a character and a number, the data is converted into character string data composed of 0s and 1s. Each of the 0s and 1s included in the character string data is treated as 1-bit data, and the 1-bit data is arranged in a data region having a two-dimensional array. Lossless compression of the data arranged in the data region is performed by using an image processing method.

Abstract: Technologies for conversion and transmission of data files from a mainframe computing system for big data ingestion. Secure electronic payment transactional data is extracted in a mainframe format, such as Extended Binary Coded Decimal Interchange Code (EBCDIC). The electronic payment transactional data is converted to American Standard Code for Information Interchange (ASCII) code prior to transmission to a big data analytics computing system. The converted data is encrypting and transmitted for ingestion into a big data analytics computing system.

Abstract: A method of encoding video data into a bitstream comprising: determining from the plurality of video blocks of a video frame of a first resolution a current block and one or more reference blocks including reference samples for predicting original samples of the current block; determining predicted samples of the current block based on a downsampling scheme, the determining comprising computing samples of a first low-resolution block of a second resolution that is lower than the first resolution, and predicting samples of one or more second low-resolution blocks of the second resolution based on the samples of the first low resolution block, the samples of the first low-resolution block and the one or more second low-resolution blocks defining predicted samples of the current Block; and, determining residual samples of a residual block based on original samples of the current block and the predicted samples of the current block; transforming the samples of the residual block into a bitstream; and, embedding m

Abstract: An arithmetic encoder for encoding a plurality of symbols having symbol values is configured to derive an interval size information for an arithmetic encoding of one or more symbol values to be encoded based on a plurality of state variable values representing statistics of a plurality of previously encoded symbol values with different adaptation time constants. The arithmetic encoder is configured to map a first state variable value, or a scaled and/or rounded version thereof, using a lookup-table and to map a second state variable value, or a scaled and/or rounded version thereof using the lookup-table, in order to obtain the interval size information describing an interval size for the arithmetic encoding of one or more symbols to be encoded. Further arithmetic encoders, arithmetic decoders, video encoders, video decoder, methods for encoding, methods for decoding and computer programs are also disclosed which are based on the same concept and on other concepts.

Abstract: The present disclosure pertains generally to image feature extraction. Both transfer-learning and multi-task training approaches are considered. In one example, a machine learning model is trained to perform a geographic classification task of distinguishing between images captured in different geographic regions based on their visual content. In another example, a machine learning model is trained to perform an order recognition task of determining information about the order of an image sequence based on its visual content, where the order of the image sequence may be different than the order in which its constituent images were captured. A further example combines the two approaches. The knowledge gained by the ML model in learning one or more such tasks can be applied to a desired image recognition task, such as image segmentation, structure detection or image classification, e.g. with a pre-training/fine-tuning framework or a multi-task learning framework.

Abstract: Aspects of the disclosure provide methods and apparatuses for point cloud compression and decompression. In some examples, an apparatus for point cloud compression/decompression includes processing circuitry. For example, the processing circuitry in the apparatus for point cloud encoding receives an occupancy map for a point cloud. The occupancy map is indicative of a background portion and a foreground portion for a coding block in an image that is generated based on the point cloud. Then, the processing circuitry devaluates distortions in the background portion of the coding block during an optimization process that results a coding option for the coding block, and encodes the coding block according to the coding option.

Abstract: The invention introduces a method for compressing texture tiles, which contains at least the following steps: lossless-compressing raw data of a texture tile; determining whether a length of the lossless-compression result of the raw data is greater than a target result; and when the length of the lossless-compression result of the raw data is greater than the target length, performing data-reduction control in layers for generating reduced data by reducing the raw data, and generating a lossless-compression result of the reduced data, thereby enabling the length of the lossless-compression result of the reduced data to be equal to the target length or shorter.