Abstract: A method of adaptive neural image compression with a hyperprior model by meta-learning is performed by at least one processor and includes generating a statistic feature, based on an input image and a hyperparameter, and generating a first shared feature and an estimated adaptive encoding parameter, encoding the input image to obtain a signal encoded image, based on the generated first shared feature and the generated estimated adaptive encoding parameter, generating a second shared feature and an estimated adaptive hyper encoding parameter, generating a hyper feature, based on the signal encoded image, the generated second shared feature, and the generated estimated adaptive hyper encoding parameter, and compressing the obtained signal encoded image, the generated statistic feature, and the generated hyper feature.

Abstract: Disclosed are three-dimensional (“3D”) graphical user interface (“GUI”) elements for improving user interactions with a digital environment or a device by simplifying access to different data, functionality, and operations of the digital environment or the device. A 3D GUI element may include first visual information at a first position and second visual information at a second position within the 3D space represented by the 3D GUI element. In response to first input directed to the first visual information, the 3D GUI or system may perform a first action that is mapped to the first input and the first visual information within the 3D GUI element. In response to second input directed to the second visual information, the 3D GUI or system may perform a second action that is mapped to the second input and the second visual information within the 3D GUI element.

Abstract: Methods and apparatuses for watermark embedding and extracting are provided. A method for watermark embedding includes obtaining a carrier object and watermark information to be embedded in the carrier object; generating at least one encoding region including the watermark information according to the watermark information, the at least one encoding region including a plurality of template lattices; obtaining a watermark image according to the at least one encoding region; and embedding the watermark image in the carrier object.

Abstract: Example method includes: transmitting a plurality of probe images from an IoT device at an edge network to a server hosting a target DNN, wherein the plurality of images are injected with a limited amount of noise to probe sensitivities of the target DNN to the red, green, and blue colors; receiving a feedback comprising a plurality of DCT coefficients unique to target DNN from the server hosting the target DNN; computing a plurality of color conversion weights based on the feedback received from the server; converting a set of real-time images from RGB color space to YUV color space using the plurality of color conversion weights unique to the target DNN; compressing the set of real-time images using a quantization table unique to the target DNN by the IoT device; and transmitting the compressed set of real-time images to the server hosting the target DNN for DNN inferences.

Abstract: A mechanism is described for facilitating power extension service at computing devices according to one embodiment of the invention. A method of embodiments of the invention includes calculating potential power saving by one or more of a plurality of power-saving techniques supported by a computing device. The calculating includes identifying the one or more of the plurality of power-saving techniques that are available for selection and an expected amount of power to be saved with the one or more of the plurality of power saving techniques. The method may further include generating a list identifying the one or more of the plurality of power-saving techniques and relevant information resulting from the calculation, and displaying the list.

Abstract: A computing device includes a memory, a display for displaying a representation of data stored in the memory, and a surface for detecting user input. In response to detecting an indication of a command, the computing device is configured to identify the command, and automatically identify one or more variables, associated with at least one function, representing a plurality of at least one location in the memory, based on at least one portion of the command. The at least one function is configured to automatically determine one of the plurality of at least one location at which to apply the command to at least one text character or graphic object, based on the one or more variables.

Abstract: Video coding systems or apparatus utilizing context-based adaptive binary arithmetic coding (CABAC) during encoding and/or decoding, are configured according to the invention with an enhanced binarization of non-zero Delta-QP (dQP). During binarization the value of dQP and the sign are separately encoded using unary coding and then combined into a binary string which also contains the dQP non-zero flag. This invention capitalizes on the statistical symmetry of positive and negative values of dQP and results in saving bits and thus a higher coding efficiency.

Abstract: A first image to be processed is identified, where the first image includes one or more interference factors. The one or more interference factors are removed from the first image using a plurality of different interference factor removal techniques to obtain a plurality of sample images, where each of the plurality of sample images is associated with a particular interference factor removal technique. Each sample image of the plurality of sample images is segmented into a plurality of sample sub-images based on a segmentation rule, where each sample sub-image is associated with an attribute. A plurality of target sub-images is determined from the plurality of sample su b-images, where each target sub-image comprises a combination of sample sub-images associated with a common attribute, and where each target sub-image is associated with a different attribute. The plurality of target sub-images associated with different attributes is combined into a target image.

Abstract: A first image to be processed is identified, where the first image includes one or more interference factors. The one or more interference factors are removed from the first image using a plurality of different interference factor removal techniques to obtain a plurality of sample images, where each of the plurality of sample images is associated with a particular interference factor removal technique. Each sample image of the plurality of sample images is segmented into a plurality of sample sub-images based on a segmentation rule, where each sample sub-image is associated with an attribute. A plurality of target sub-images is determined from the plurality of sample sub-images, where each target sub-image comprises a combination of sample sub-images associated with a common attribute, and where each target sub-image is associated with a different attribute. The plurality of target sub-images associated with different attributes is combined into a target image.

Abstract: Morphological anti-ringing is disclosed. A method for encoding a block of an image includes generating a transform block for the block of first values, generating quantized transform coefficients for the transform block, determining a clamping value for the block, encoding the quantized transform coefficients in a compressed bitstream, and encoding the clamping value in the compressed bitstream. The clamping value is used to clamp second values. The second values correspond to the first values and result from inverse transforming the transform block. An apparatus for decoding a block of an image includes a memory and a processor configured to execute instructions stored in the memory to decode, from a compressed bitstream, a quantized transform block; obtain, using the quantized transform block, a decoded block of pixel values; decode, from a compressed bitstream, a clamping value for the block; and clamp the pixel values using the clamping value.

Abstract: A compression system trains a machine-learned encoder and decoder. The encoder can be deployed by a sender system to encode content for transmission to a receiver system, and the decoder can be deployed by the receiver system to decode the encoded content and reconstruct the original content. The encoder receives content and generates a tensor as a compact representation of the content. The content may be, for example, images, videos, or text. The decoder receives a tensor and generates a reconstructed version of the content. In one embodiment, the compression system trains one or more encoding components such that the encoder can adaptively encode different degrees of information for regions in the content that are associated with characteristic objects, such as human faces, texts, or buildings.

Abstract: The invention relates to a method for compressing a set of input binary data values x, all coded in a same number B of bits, into a corresponding set of output data values x?, all coded in a smaller number b of bits, obtainable by (i) computing a quantization step size dq dq = 2 B + 1 2 b + 1 - 1 (ii) dividing the range of values [0, 2B?1[, into 2b subranges, the first subrange being [0, INT(dq/2)[ the ith subrange being [INT((i?3/2)*dq)+1, INT((i?1/2)*dq)[, for i going from 2 to 2b The invention also relates to a method for decompressing data and to applications of said method for compressing/decompressing video data, and to devices for performing these operations.

Abstract: The bit-rate of surveillance video streams can be adapted by varying the number of enhancement layers/streams that are transmitted to a local monitoring center and/or data storage center. Enhancement layers that are not transmitted to the local data/monitoring center may be sent to a remote storage location, thereby allowing users to reproduce a high quality version of the surveillance video by retrieving those enhancement layers from the remote storage location. The base and enhancement layers may be obtained by encoding the surveillance video in accordance with a scalable video coding format, such as the H.264/MPEG-4 Advanced video coding (AVC) codec.

Abstract: Decoder processor hardware reproduces a support plane including a set of support values. The set of support values is derived from combining a sequence of multiple original sets of values. The decoder processor hardware receives compensation information. The decoder processor hardware utilizes the compensation information to produce preliminary renditions of the multiple original sets of values based on the set of support values. Encoder processor hardware processes a sequence of original sets of values. The encoder processor hardware utilizes the values of the original sets in the sequence to produce a set of support values, the set of support values representing a baseline to reproduce a rendition of each of the original sets of values. The encoder processor hardware generates reconstruction data to include data corresponding to the set of support values, the reconstruction data indicates how to reconstruct the original sets of values using the set of support values.

Abstract: In an example, system and method for coding encoded video data. A quantized residue differences block associated with a current block of a picture is generated and a reconstructed prediction residue for each quantized residue difference in the block of quantized residue differences is generated. Each reconstructed prediction residue is added to its corresponding original prediction value to produce a current block of a picture. Generating the reconstructed prediction residue includes performing inverse quantization on blocks of quantized residue differences, wherein the inverse quantization reconstructs data to which uniform quantization residue differential pulse code modulation (RDPCM) has been applied and wherein the quantization is of the form: ?Xq=floor((X+?Q)/Q) where Q is quantization step and ? is quantization offset.

Abstract: An image processing method includes receiving a bitstream comprising a first encoding unit and a second encoding unit; acquiring from the bitstream a reference value for decoding of the first coding unit, a weight value for performing interpolation using the reference value to determine one or more sample values, and an index indicating one of the one or more sample values; decoding the first encoding unit using the index and a result of interpolation performed using the reference value and the weight value; and decoding the second encoding unit from the bitstream using the interpolation result used in the decoding of the first encoding unit.

Abstract: There is provided an information processing apparatus. A multidimensional input vector is input. For each dimension of the input vector, a function value of a single-variable function with an element of the dimension as a variable is derived, by referring to a lookup table indicating a correspondence between a variable and a function value of the single-variable function. A product of the single-variable functions approximates a function value of a multiple-variable function. For each dimension of the input vector, a product of the function value derived by the derivation unit and a predetermined coefficient corresponding to the dimension is calculated. A value calculated using the total of the products calculated by the product calculation unit for each dimension of the input vector is output as a classification index indicating a class of the input vector.

Abstract: A system and method for modifying media stream based on a channel limitation are disclosed. In one embodiment, the system includes a channel analyzer, a modality of interest (MOI) identification module and an inter-modality resolution modification module. The channel analyzer determines a channel limitation. The MOI identification module receives a plurality of media streams and identifies at least one MOI. The inter-modality resolution modification module modifies a resolution of at least one of the media streams based on the channel limitation and content of the media streams. In another embodiment, the system includes a region of interest identification (ROI) module and an intra-modality resolution modification module instead of, or in addition to, the MOI identification module and inter-modality resolution modification module. The ROI identification module receives a media stream and identifies at least one ROI within the media stream.

Abstract: In one example embodiment, a method performed under control of a server includes: receiving a user video from a transmitting device; generating conference video data that includes at least one portion of the user video; assigning, in accordance with a security level of a receiving device, a first portion of the conference video data a first security attribute; generating first security data corresponding to the first security attribute; and streaming, to the receiving device, the conference video data that includes the at least one portion of the user video and the first security data.

Abstract: An image processing system and method of operation includes: a source image having source pixels; a homogeneity image module for forming a homogeneity image having homogeneity pixels each with a homogeneity value indicating the local homogeneity of source pixels of the source image; a watershed image module, coupled to the homogeneity image module, for forming a watershed image having watershed segments with a segment weight, the watershed image formed by segmenting the homogeneity image, and for merging one of the watershed segments with another of the watershed segments both having the segment weight less than or equal to a segment threshold; a contour map module, coupled to the watershed image module, for forming a segmentation contour map having boundaries of the watershed segments formed by merging the watershed segments; and a composite image module for forming a composite image with the segmentation contour map for display on a device.

Abstract: Methods and apparatus for converting original data into a plurality of sub-bands using wavelet decomposition; encrypting at least one of the sub-bands using a key to produce encrypted sub-band data; and transmitting the encrypted sub-band data to a recipient separately from the other sub-bands.

Abstract: At the transmitter side, carrier waves are modulated according to an input signal for producing relevant signal points in a signal space diagram. The input signal is divided into two, first and second, data streams. The signal points are divided into signal point groups to which data of the first data stream are assigned. Also, data of the second data stream are assigned to the signal points of each signal point group. A difference in the transmission error rate between first and second data streams is developed by shifting the signal points to other positions in the space diagram expressed at least in the polar coordinate system. At the receiver side, the first and/or second data streams can be reconstructed from a received signal. In TV broadcast service, a TV signal is divided by a transmitter into, low and high, frequency band components which are designated as a first and a second data streams respectively.

Abstract: At the transmitter side, carrier waves are modulated according to an input signal for producing relevant signal points in a signal space diagram. The input signal is divided into, two, first and second, data streams. The signal points are divided into signal point groups to which data of the first data stream are assigned. Also, data of the second data stream are assigned to the signal points of each signal point group. A difference in the transmission error rate between first and second data streams is developed by shifting the signal points to other positions in the space diagram expressed at least in the polar coordinate system. At the receiver side, the first and/or second data streams can be reconstructed from a received signal. In TV broadcast service, a TV signal is divided by a transmitter into, low and high, frequency band components which are designated as a first and a second data streams respectively.

Abstract: At the transmitter side, carrier waves are modulated according to an input signal for producing relevant signal points in a signal space diagram. The input signal is divided into, two, first and second, data streams. The signal points are divided into signal point groups to which data of the first data stream are assigned. Also, data of the second data stream are assigned to the signal points of each signal point group. A difference in the transmission error rate between first and second data streams is developed by shifting the signal points to other positions in the space diagram expressed at least in the polar coordinate system. At the receiver side, the first and/or second data streams can be reconstructed from a received signal. In TV broadcast service, a TV signal is divided by a transmitter into, low and high, frequency band components which are designated as a first and a second data streams respectively.

Abstract: At the transmitter side, carrier waves are modulated according to an input signal for producing relevant signal points in a signal space diagram. The input signal is divided into, two, first and second, data streams. The signal points are divided into signal point groups to which data of the first data stream are assigned. Also, data of the second data stream are assigned to the signal points of each signal point group. A difference in the transmission error rate between first and second data streams is developed by shifting the signal points to other positions in the space diagram expressed at least in the polar coordinate system. At the receiver side, the first and/or second data streams can be reconstructed from a received signal. In TV broadcast service, a TV signal is divided by a transmitter into, low and high, frequency band components which are designated as a first and a second data streams respectively.