Abstract: An information processing apparatus obtains first feature data generated from image data including a plurality of pixels given different first coordinates, the first feature data including a plurality of feature values given different second coordinates. The information processing apparatus generates a first inference result indicating an image region of the image data by feeding the first feature data to a first machine learning model. The information processing apparatus generates second feature data corresponding to the image region from the first feature data on the basis of coordinate mapping information indicating mapping between the first coordinates and the second coordinates. The information processing apparatus generates a second inference result for the image region by feeding the second feature data to a second machine learning model.

Abstract: An image processing system includes: devices that obtain inputted images; servers that perform an inference process on the inputted images; and a controlling apparatus that controls the devices and the servers. A first device obtains a first feature of a first image by inputting the first image into a former-part layer of a machine learning model that performs the inference process, calculates statistics information of the first feature and transmits to the controlling apparatus. The controlling apparatus determines a network band and a first server based on the statistics information and performance of each server, the network band being allocated to the first device. The first device transmits the first feature to the first server based on the network band. The first server obtains an inference result by inputting the first feature received from the first device into a latter-part layer of the machine learning model.

Abstract: An information processing system includes an edge computer that implements a preceding stage of a learning model, and a cloud computer that implements a subsequent stage of the learning model, wherein the edge computer includes a first processor configured to calculate a first feature amount by inputting a first image to the preceding stage, identify an area of interest in the first image based on the first feature amount, generate a second image obtained by masking the area of interest in the first image, calculate a second feature amount by inputting the second image to the preceding stage, and transmit the second feature amount to the cloud computer, and the cloud computer includes a second processor configured to infer an object included in the second image by inputting the second feature amount to the subsequent stage.

Abstract: An image processing device includes: a memory; and a processor coupled to the memory and configured to: calculate a degree of influence of each pixel of image data, the influence being exerted on a processing result when the image data is input to a deep learning model; reduce an information amount of intermediate information extracted from the deep learning model based on the degree of influence; and compress the intermediate information, the information amount of which has been reduced.

Abstract: An image recognition system includes at least one memory, and at least one processor coupled to the at least one memory, respectively, and configured to perform a first process, for an image inputted, up to a dividing position determined in a DNN, and output feature-maps of the image, compress the outputted feature-maps, and transmit the compressed feature-maps, receive and reconstruct the compressed feature-maps, and perform a recognition process, for the reconstructed feature-map inputted, after the dividing position, and output a recognition result, wherein the dividing position is determined by accuracy of the recognition result and a total time that includes a time for the performing of the first process, a time for the compressing of the outputted feature-maps, a time for the transmitting of the compressed feature-maps, a time for the reconstructing of the compressed feature-maps, and a time for the performing of the recognition process.

Abstract: An image processing apparatus, includes a memory; and a processor coupled to the memory and the processor configured to: identify a first recognition error, the first recognition error being an error between ground truth data and a first recognition result obtained by inputting a first feature of image data into a first image recognition model, generate a second feature obtained by adding noise to the first feature of the image data, identify a second recognition error, the second recognition error being an error between the first recognition result and a recognition result obtained by inputting the second feature into a second image recognition model, and execute training of the first image recognition model and the second image recognition model based on the first recognition error and the second recognition error.

Abstract: An image recognition system includes at least a memory, and at least a processor coupled to at least the memory, respectively, and configured to extract feature maps from an input image, compress the extracted feature maps, reconstruct the compressed feature maps, reconstruct an image from the reconstructed feature maps and output the reconstructed image, and recognize the input image based on the reconstructed feature maps and output a recognition result, wherein the compressing the extracted feature maps and the reconstructing the compressed feature maps are learned so as to minimize a cost based on an information amount when compressing the extracted feature maps, a first error between the input image and the reconstructed image, and a second error between the recognition result and a ground truth.

Abstract: An image processing apparatus, includes a memory; and a processor coupled to the memory and the processor configured to: Identify a first recognition error, the first recognition error being an error between ground truth data and a first recognition result obtained by inputting a first feature of image data into a first image recognition model, generate a second feature obtained by adding noise to the first feature of the image data, identify a second recognition error, the second recognition error being an error between the first recognition result and a recognition result obtained by inputting the second feature into a second image recognition model, and execute training of the first image recognition model and the second image recognition model based on the first recognition error and the second recognition error.

Abstract: An information processing apparatus includes: a memory; and a processor coupled to the memory and configured to: down-convert a moving image in units of structure of pictures (SOPs) divided in a time direction and precedingly execute encoding processing; calculate a target code amount of each processing unit included in the moving image in units of SOPs based on a result of the preceding encoding processing and calculate a temporal buffer position in a case where the encoding processing is executed on each processing unit based on the calculated target code amount; and calculate an error between an actual transmission buffer position and a temporal buffer position each time when the encoding processing is executed on each processing unit in the moving image in units of SOPs and correct a target code amount of each processing unit on which the encoding processing has not been executed yet.

Abstract: A video coding device includes a memory and a processor coupled to the memory. The processor is configured to sequentially receive a plurality of pictures and detect a scene change based on the plurality of pictures. When a scene change is detected, the processor determines whether a given condition is satisfied by a position of a first picture where the scene change is detected. When the given condition is satisfied by the position of the first picture, the processor redisplays, instead of the first picture, a forward second picture among a plurality of pictures referenced by the first picture.

Abstract: An apparatus includes: a processor configured to: execute temporal-scalable coding on input-video-image data at a first frame rate; convert the input-video-image data to video-image data of a second frame rate lower than the first frame rate; execute temporal-scalable coding on the video-image data of the second frame-rate and output a second encoding-complexity measure of a picture subjected to the temporal-scalable coding at the second frame-rate; correct, when the temporal-scalable coding is executed on the input-video-image data at the first frame-rate, a first encoding-complexity measure, estimated based on the second encoding-complexity of the picture and to be referenced for a picture to be encoded, of the picture to be encoded, based on a distance between the picture to be encoded and the picture to be referenced in a time direction; and execute the temporal-scalable coding on the input-video-image data at the first frame-rate based on the corrected first encoding-complexity measure.

Abstract: An information processing apparatus includes: a memory; and a processor coupled to the memory and configured to: down-convert a moving image in units of structure of pictures (SOPs) divided in a time direction and precedingly execute encoding processing; calculate a target code amount of each processing unit included in the moving image in units of SOPs based on a result of the preceding encoding processing and calculate a temporal buffer position in a case where the encoding processing is executed on each processing unit based on the calculated target code amount; and calculate an error between an actual transmission buffer position and a temporal buffer position each time when the encoding processing is executed on each processing unit in the moving image in units of SOPs and correct a target code amount of each processing unit on which the encoding processing has not been executed yet.

Abstract: A video coding device includes a memory and a processor coupled to the memory. The processor is configured to sequentially receive a plurality of pictures and detect a scene change based on the plurality of pictures. When a scene change is detected, the processor determines whether a given condition is satisfied by a position of a first picture where the scene change is detected. When the given condition is satisfied by the position of the first picture, the processor redisplays, instead of the first picture, a forward second picture among a plurality of pictures referenced by the first picture.

Abstract: An encoder includes: a plurality of first processors; and a second processor configured to: generate reduced image information by reducing image information; determine that a first block is a preferential object block when it is determined, based on a direction of a motion vector of the first block, that the first block is a block to be encoded with reference to a block included in a second reduced slice adjacent to a first reduced slice among reduced slices obtained by dividing the reduced image information, the first block being included in the first reduced slice; and perform, when the first block is a preferential object block, a control to reduce a first quantization parameter used by one of the plurality of first processors to encode a block corresponding to the first block among a plurality of blocks included in a first slice corresponding to the first reduced slice.

Abstract: An apparatus includes: a processor configured to: execute temporal-scalable coding on input-video-image data at a first frame rate; convert the input-video-image data to video-image data of a second frame rate lower than the first frame rate; execute temporal-scalable coding on the video-image data of the second frame-rate and output a second encoding-complexity measure of a picture subjected to the temporal-scalable coding at the second frame-rate; correct, when the temporal-scalable coding is executed on the input-video-image data at the first frame-rate, a first encoding-complexity measure, estimated based on the second encoding-complexity of the picture and to be referenced for a picture to be encoded, of the picture to be encoded, based on a distance between the picture to be encoded and the picture to be referenced in a time direction; and execute the temporal-scalable coding on the input-video-image data at the first frame-rate based on the corrected first encoding-complexity measure.

Abstract: A moving image encoding device includes determining whether an orthogonal transformation coefficient to which orthogonal transformation processing has been executed is equal to or less than a threshold determined from a quantization step used for quantization processing, and first calculating an encoding error from a first coefficient determined to be the threshold or less among the orthogonal transformation coefficients and the number of second coefficients determined to be larger than the threshold, and estimating, from the number of the second coefficients, an information amount generated when variable length encoding processing is executed to the orthogonal transformation coefficient after execution of the quantization processing, and second calculating an encoding cost from the encoding error and the information amount.

Abstract: An apparatus for encoding a moving picture data by an intra-refresh technique includes a memory, and a processor coupled to the memory. The apparatus performs setting of a width of an intra slice in each of a plurality of pictures included in the moving picture data within a refresh cycle having a given length such that a generated information amount in the intra slice is equalized between each of the plurality of pictures. The apparatus performs an intra-refresh to a region corresponding to the intra slice. The apparatus performs encoding of each block included in a refreshed region, through which the intra slice is passed in the refresh cycle, by referring to an encoded pixel in the refreshed region of the picture or in the refreshed region of an encoded picture whose encoding sequential order is before that of the picture.

Abstract: A video coding apparatus includes a processor that executes a procedure. The procedure includes: (c) classifying each of a plurality of groups into either a full evaluation group or into a partial evaluation group; (d) determining the prediction mode to use as generating the predicted image of a coding target block by evaluating coding costs of all the prediction modes belonging to each of the groups for a group classified at (c) into the full evaluation group, and evaluating coding costs of a portion of prediction modes out of the prediction modes belonging to each of the groups for a group classified at (c) into the partial evaluation group; and (e) generating a predicted image of the coding target block using the prediction mode determined at (d).

Abstract: A moving image encoding apparatus configured to use an image one frame before a target image for encoding to calculate a filtering coefficient for each intra prediction mode of a decoded image from the decoded image and the original image, extracts a filtering coefficient to perform a filtering process to a reference pixel for intra prediction mode determination of a peripheral block of the target block for encoding according to the intra prediction mode of the peripheral block and to generate a reference image for intra prediction mode determination, uses the reference image for intra prediction mode determination to select an intra prediction mode for the target block for encoding, and generates an intra prediction image of the target block for encoding.

Abstract: A moving image encoding device includes determining whether an orthogonal transformation coefficient to which orthogonal transformation processing has been executed is equal to or less than a threshold determined from a quantization step used for quantization processing, and first calculating an encoding error from a first coefficient determined to be the threshold or less among the orthogonal transformation coefficients and the number of second coefficients determined to be larger than the threshold, and estimating, from the number of the second coefficients, an information amount generated when variable length encoding processing is executed to the orthogonal transformation coefficient after execution of the quantization processing, and second calculating an encoding cost from the encoding error and the information amount.