Abstract: There is disclosed an apparatus, a method and a computer program for video coding. The apparatus comprises a selector configured for selecting a pixel for prediction; a projection definer configured for determining a projection of said pixel to a set of reference pixels; and a prediction definer configured for selecting one or more reference pixels from said set of reference pixels on the basis of said projection, and using said selected one or more reference pixels to obtain a prediction value for said pixel to be predicted.

Abstract: An apparatus includes circuitry configured to: partition an input tensor into one or more block tensors; partition at least one of the block tensors into one or more continuation bands, the one or more continuation bands being associated with a caching counter having a value; store the one or more continuation bands in a cache managed using a cache manager; retrieve, prior to a convolution or pooling operation on a current block tensor, the one or more continuation bands of a previous block tensor from the cache that are adjacent to a current block tensor; concatenate the retrieved continuation bands with the current block tensor; apply the convolution or pooling operation on the current block tensor after the concatenation; decrease the respective caching counter value of the retrieved continuation bands; and clear the continuation bands from the cache when its respective caching counter reaches a value of zero.

Abstract: Video data is obtained or received. At least a current frame or previous frame(s) of the obtained or received video data are provided to an input of a neural network. A predicted output is generated at an output of the neural network. The predicted output includes at least one of predicted future frame(s) of the video data and predicted properties of future frame(s) of the video data. Processing decision(s) are determined based, at least in part, on the predicted output. The current frame of the video data is processed at least partially according to the processing decision(s).

Abstract: A method includes maintaining a set of parameters or weights derived through online learning for a neural net; transmitting an update of the parameters or weights to a decoder; deriving a first prediction block based on an output of the neural net using the parameters or weights; deriving a first encoded prediction error block through encoding a difference of the first prediction block and a first input block; encoding the first encoded prediction error block into a bitstream; deriving a reconstructed prediction error block based on the first encoded prediction error block; deriving a second prediction block based on an output of the neural net using the parameters or weights and the reconstructed prediction error block; deriving a second encoded prediction error block through encoding a difference of the second prediction block and a second input block; and encoding the second encoded prediction error block into a bitstream.

Abstract: There are disclosed various methods, apparatuses and computer program products for video decoding or encoding. In some embodiments for decoding a block that contains quantized residual coefficients based on at least two color channels of a video presentation or an image is received. A first quantization parameter and an offset defining a relationship between the first quantization parameter and a second quantization parameter for the block are obtained. The value of the second quantization parameter is determined by using the first quantization parameter and the offset. The residual samples are reconstructed by dequantizing the quantized residual coefficients using the second quantization parameter, if the block has been encoded by using a cross-channel residual coding mode.

Abstract: A method for motion compensated prediction, the method comprising determining a motion vector for a block of samples; determining a sub-sample accurate horizontal component and a sub-sample accurate vertical component of said motion vector;determining fractional parts of said sub-sample accurate horizontal and vertical motion vector components; determining interpolation filter length and interpolation filter based on said fractional parts; applying said interpolation filter with determined length to perform a filtering operation at least in either horizontal or vertical direction; and storing the result of said filtering operation as the motion compensated prediction with said motion vector.

Abstract: The embodiments relate to a method comprising determining a coding mode of a transform block, wherein a transform block comprises a set of transform coefficients; determining a shape of the transform block; determining at least one transform mode for a block based at least partly on said coding mode and said shape of the transform block; applying the determined transform mode to a set of transform coefficients to produce sample values; and adding said sample values to a block of predicted sample values. The embodiments also relate to technical equipment for implementing the method.

Abstract: A method comprising: deriving a first prediction block (608) at least partly based on an output of a neural net (602) using a first set of parameters; deriving a first encoded prediction error block (614-620) through encoding a difference of the first prediction block and a first input block; encoding (620) the first encoded prediction error block into a bitstream; deriving a first reconstructed prediction error block (624) from the first encoded prediction error block; deriving a training signal (628) from one or both of the first encoded prediction error block and/or the first reconstructed prediction error block (624); retraining (630) the neural net (602) with the training signal (628) to obtain a second set of parameters for the neural net (602); deriving a second prediction block (608) at least partly based on an output of the neural net using the second set of parameters; deriving a second encoded prediction error block (614-620) through encoding a difference of the second prediction block and a second

Abstract: There are disclosed various methods, apparatuses and computer program products for video decoding or encoding. In some embodiments the method for decoding or encoding comprises obtaining absolute values of at least a first transform coefficient and a second transform coefficient of a block of a picture (600); determining the sign of at least the first transform coefficient in the block (602); determining a reference measure based on at least said absolute value of the first transform coefficient and the determined sign of the first transform coefficient (604); determining a predicted sign for said second transform coefficient in the block (606); and decoding or encoding the sign of said second transform coefficient based on the determined predicted sign of said second transform coefficient (608).

Abstract: A method and apparatus are provided for encoding and decoding image information. The encoding comprises receiving a block of pixels; creating a set of motion vector prediction candidates for the block of pixels; and examining the set to determine if a motion vector prediction candidate is a temporal motion vector prediction, or a spatial motion vector prediction. If the motion vector prediction candidate is a temporal motion vector prediction, the motion vector prediction candidate is kept in the set. If the motion vector prediction candidate is a spatial motion vector prediction, it is examined whether the set comprises a motion vector prediction candidate corresponding with the spatial motion vector prediction; and if so, the motion vector prediction candidate is removed from the set. Once the set is created, one of the candidates from the set is selected to represent a motion vector prediction for the block of pixels.

Abstract: A method for motion compensated prediction, the method comprising determining a residual signal for at least one sample; determining if said residual signal is representing residual for samples in more than one channel; and if affirmative, applying said residual signal for at least a first sample in a first channel for generating a first reconstructed sample; and applying said residual signal for at least a second sample in a second channel for generating a second reconstructed sample.

Abstract: A method and apparatus are provided for encoding and decoding image information. The encoding comprises receiving a block of pixels; creating a set of motion vector prediction candidates for the block of pixels; and examining the set to determine if a motion vector prediction candidate is a temporal motion vector prediction, or a spatial motion vector prediction. If the motion vector prediction candidate is a temporal motion vector prediction, the motion vector prediction candidate is kept in the set. If the motion vector prediction candidate is a spatial motion vector prediction, it is examined whether the set comprises a motion vector prediction candidate corresponding with the spatial motion vector prediction; and if so, the motion vector prediction candidate is removed from the set. Once the set is created, one of the candidates from the set is selected to represent a motion vector prediction for the block of pixels.

Abstract: Apparatuses, methods and computer programs are provided for utilizing motion prediction in video coding. A block of pixels of a video representation encoded in a bitstream is read, and a type of the block is determined. If the determining indicates that the block is a block predicted by using two or more reference blocks, a first reference pixel location in a first reference block is determined and a second reference pixel location in a second reference block is determined. The first reference pixel location is used to obtain a first prediction. The first prediction has a second precision, which is higher than the first precision. The second reference pixel location is used to obtain a second prediction, which also has the second precision. The first prediction and the second prediction are combined to obtain a combined prediction; and the precision of the combined prediction is reduced to the first precision.

Abstract: A method comprising: deriving a first prediction block (608) at least partly based on an output of a neural net (602) using a first set of parameters; deriving a first encoded prediction error block (614-620) through encoding a difference of the first prediction block and a first input block; encoding (620) the first encoded prediction error block into a bitstream; deriving a first reconstructed prediction error block (624) from the first encoded prediction error block; deriving a training signal (628) from one or both of the first encoded prediction error block and/or the first reconstructed prediction error block (624); retraining (630) the neural net (602) with the training signal (628) to obtain a second set of parameters for the neural net (602); deriving a second prediction block (608) at least partly based on an output of the neural net using the second set of parameters; deriving a second encoded prediction error block (614-620) through encoding a difference of the second prediction block and a second

Abstract: The invention relates to a method for decoding, comprising receiving a bitstream comprising encoded image data, forming from the bitstream in a computer memory a set of reference pixels having reference pixel values for intra-picture directional prediction, predicting a pixel value in a predicted image block (P) by intra-picture directional prediction by using values of one or more selected reference pixels (410, 420) in forming a predicted a pixel value, wherein the selected reference pixels (410, 420) are selected from a prediction direction with respect to the predicted pixel, and wherein the prediction direction between the reference pixel (410) and the predicted pixel can be a wide-angle prediction direction that forms an obtuse angle with top-left direction, repeating the pixel prediction for a number of pixels to form a predicted image block, and using the predicted image block (P) in decoding an image block from the bitstream to obtain a decoded image block.

Abstract: An apparatus, a method and a computer program product are described comprising: obtaining or receiving video data; providing a current frame and/or one or more previous frames of the obtained or received video data to an input of a neural network; generating a predicted output at an output of the neural network, wherein the predicted output comprises at least one of one or more predicted future frames of the video data and predicted properties of one or more future frames of the video data; determining one or more processing decisions based, at least in part, on the predicted output; and processing the current frame of the video data at least partially according to the one or more processing decisions.

Abstract: Apparatuses, methods and computer programs are provided for utilizing motion prediction in video coding. A block of pixels of a video representation encoded in a bitstream is read, and a type of the block is determined. If the determining indicates that the block is a block predicted by using two or more reference blocks, a first reference pixel location in a first reference block is determined and a second reference pixel location in a second reference block is determined. The first reference pixel location is used to obtain a first prediction. The first prediction has a second precision, which is higher than the first precision. The second reference pixel location is used to obtain a second prediction, which also has the second precision. The first prediction and the second prediction are combined to obtain a combined prediction; and the precision of the combined prediction is reduced to the first precision.

Abstract: A method comprising: obtaining a block of a picture or a picture in an encoder; determining if the block/picture is used for on-line learning; if affirmative, encoding the block/picture; reconstructing a coarse version of the block/picture or the respective prediction error block/picture; enhancing the coarse version using a neural net; fine-tuning the neural net with a training signal based on the coarse version; determining if the block/picture is enhanced using the neural net; and if affirmative, encoding the block/picture with enhancing using the neural net.

Abstract: A method comprising: decoding, from a bitstream, a first encoded region of first picture into a first preliminary reconstructed region; forming a first reconstructed region from the first preliminary reconstructed region, wherein the forming comprises resampling and/or rearranging the first preliminary reconstructed region, wherein the rearranging comprises relocating, rotating and/or mirroring; and decoding at least a second region, wherein the first reconstructed region is used as a reference for prediction in decoding the at least second region and the second region either belongs to a second picture and is spatially collocated with the first reconstructed region or belongs to the first picture.

Abstract: There are disclosed various methods, apparatuses and computer program products for video decoding or encoding. In some embodiments the method for decoding or encoding comprises obtaining absolute values of at least a first transform coefficient and a second transform coefficient of a block of a picture (600); determining the sign of at least the first transform coefficient in the block (602); determining a reference measure based on at least said absolute value of the first transform coefficient and the determined sign of the first transform coefficient (604); determining a predicted sign for said second transform coefficient in the block (606); and decoding or encoding the sign of said second transform coefficient based on the determined predicted sign of said second transform coefficient (608).