Abstract: Methods and processing modules apply adaptive sharpening, for a block of input pixels, to determine a block of output pixels. A block of sharp pixels is obtained based on the block of input pixels, the block of sharp pixels being for representing a sharp version of the block of output pixels. One or more indications of contrast for the block of input pixels is determined. Each of the output pixels of the block of output pixels is determined by performing a respective weighted sum of: (i) a corresponding input pixel in the block of input pixels and (ii) a corresponding sharp pixel in the block of sharp pixels. The weights of the weighted sums are based on the determined one or more indications of contrast for the block of input pixels.

Abstract: Methods and systems for implementing a traditional computer vision algorithm as a neural network. The method includes: receiving a definition of the traditional computer vision algorithm that identifies a sequence of one or more traditional computer vision algorithm operations; mapping each of the one or more traditional computer vision algorithm operations to a set of one or more neural network primitives that is mathematically equivalent to that traditional computer vision algorithm operation; linking the one or more network primitives mapped to each traditional computer vision algorithm operation according to the sequence to form a neural network representing the traditional computer vision algorithm; and configuring hardware logic capable of implementing a neural network to implement the neural network that represents the traditional computer vision algorithm.

Abstract: Methods and processing modules apply adaptive sharpening, for a block of input pixels for which processing is performed, to determine a block of output pixels. A block of non-sharp processed pixels is obtained based on the block of input pixels, the block of non-sharp processed pixels being for representing a non-sharp version of the block of output pixels. A block of sharp processed pixels is obtained based on the block of input pixels, the block of sharp processed pixels being for representing a sharp version of the block of output pixels. One or more indications of contrast for the block of input pixels is determined. Each of the output pixels of the block of output pixels is determined by performing a respective weighted sum of: (i) a corresponding non-sharp processed pixel in the block of non-sharp processed pixels and (ii) a corresponding sharp processed pixel in the block of sharp processed pixels.

Abstract: Methods and processing modules upsample a block of input pixels to determine a block of upsampled pixels. At least one of the upsampled pixels is a diagonal pixel, wherein a diagonal pixel is at a position that is not in any of the rows nor in any of the columns of input pixels in the block of input pixels. Indications of image gradients are determined for the block of input pixels. The determined indications of image gradients are used to determine one or more weighting parameters which are indicative of weights of a diagonal kernel. The upsampled pixels of the block of upsampled pixels are determined by applying kernels to the block of input pixels, wherein the diagonal pixel in the block of upsampled pixels is determined by applying the diagonal kernel to the block of input pixels in accordance with the determined one or more weighting parameters.

Abstract: A method in a hardware implementation of a Convolutional Neural Network (CNN), includes receiving a first subset of data having at least a portion of weight data and at least a portion of input data for a CNN layer and performing, using at least one convolution engine, a convolution of the first subset of data to generate a first partial result; receiving a second subset of data comprising at least a portion of weight data and at least a portion of input data for the CNN layer and performing, using the at least one convolution engine, a convolution of the second subset of data to generate a second partial result; and combining the first partial result and the second partial result to generate at least a portion of convolved data for a layer of the CNN.

Abstract: A method of rendering an image of a 3-D scene includes rendering a noisy image; and obtaining one or more guide channels. For each of a plurality of local neighbourhoods, the method comprises: calculating the parameters of a model that approximates the noisy image as a function of the one or more guide channels, and applying the calculated parameters to produce a denoised image. Tiling is used when calculating the parameters of the model.

Abstract: Methods and systems for performing a convolution transpose operation between an input tensor having a plurality of input elements and a filter comprising a plurality of filter weights. The method includes: dividing the filter into a plurality of sub-filters; performing, using hardware logic, a convolution operation between the input tensor and each of the plurality of sub-filters to generate a plurality of sub-output tensors, each sub-output tensor comprising a plurality of output elements; and interleaving, using hardware logic, the output elements of the plurality of sub-output tensors to form a final output tensor for the convolution transpose.

Abstract: A method in a hardware implementation of a Convolutional Neural Network (CNN), includes receiving a first subset of data having at least a portion of weight data and at least a portion of input data for a CNN layer and performing, using at least one convolution engine, a convolution of the first subset of data to generate a first partial result; receiving a second subset of data comprising at least a portion of weight data and at least a portion of input data for the CNN layer and performing, using the at least one convolution engine, a convolution of the second subset of data to generate a second partial result; and combining the first partial result and the second partial result to generate at least a portion of convolved data for a layer of the CNN.

Abstract: Methods for determining a fixed point format for one or more layers of a DNN based on the portion of the output error of the DNN attributed to the fixed point formats of the different layers. Specifically, in the methods described herein the output error of a DNN attributable to the quantisation of the weights or input data values of each layer is determined using a Taylor approximation and the fixed point number format of one or more layers is adjusted based on the attribution. For example, where the fixed point number formats used by a DNN comprises an exponent and a mantissa bit length, the mantissa bit length of the layer allocated the lowest portion of the output error may be reduced, or the mantissa bit length of the layer allocated the highest portion of the output error may be increased. Such a method may be iteratively repeated to determine an optimum set of fixed point number formats for the layers of a DNN.

Abstract: Methods and systems for performing a convolution transpose operation between an input tensor having a plurality of input elements and a filter comprising a plurality of filter weights. The method includes: dividing the filter into a plurality of sub-filters; performing, using hardware logic, a convolution operation between the input tensor and each of the plurality of sub-filters to generate a plurality of sub-output tensors, each sub-output tensor comprising a plurality of output elements; and interleaving, using hardware logic, the output elements of the plurality of sub-output tensors to form a final output tensor for the convolution transpose.

Abstract: A histogram-based method of selecting a fixed point number format for representing a set of values input to, or output from, a layer of a Deep Neural Network (DNN). The method comprises obtaining a histogram that represents an expected distribution of the set of values of the layer, each bin of the histogram is associated with a frequency value and a representative value in a floating point number format; quantising the representative values according to each of a plurality of potential fixed point number formats; estimating, for each of the plurality of potential fixed point number formats, the total quantisation error based on the frequency values of the histogram and a distance value for each bin that is based on the quantisation of the representative value for that bin; and selecting the fixed point number format associated with the smallest estimated total quantisation error as the optimum fixed point number format for representing the set of values of the layer.

Abstract: A method of rendering an image of a 3-D scene includes rendering a noisy image; and obtaining one or more guide channels. For each of a plurality of local neighbourhoods, the method comprises: calculating the parameters of a model that approximates the noisy image as a function of the one or more guide channels, and applying the calculated parameters to produce a denoised image. Tiling is used when calculating the parameters of the model.

Abstract: Methods and systems for implementing a traditional computer vision algorithm as a neural network. The method includes: receiving a definition of the traditional computer vision algorithm that identifies a sequence of one or more traditional computer vision algorithm operations; mapping each of the one or more traditional computer vision algorithm operations to a set of one or more neural network primitives that is mathematically equivalent to that traditional computer vision algorithm operation; linking the one or more network primitives mapped to each traditional computer vision algorithm operation according to the sequence to form a neural network representing the traditional computer vision algorithm; and configuring hardware logic capable of implementing a neural network to implement the neural network that represents the traditional computer vision algorithm.

Abstract: A mechanism for performing a discrete Fourier-related transform using a hardware accelerator that comprises fixed-function circuitry including convolution hardware configured to perform one or more convolution operations. A matrix multiplication operation used in the discrete Fourier-related transform is performed by the convolution hardware using a convolution operation. A convolution kernel for the convolution operation is derived from a weight matrix representing a multiplicand or multiplier of the matrix multiplication operation.

Abstract: A method of rendering an image of a 3-D scene includes rendering a noisy image at a first resolution; obtaining one or more guide channels at the first resolution, and obtaining one or more corresponding guide channels at a second resolution. The second resolution may be the same resolution as, or a higher resolution than, the first resolution. For each of a plurality of local neighbourhoods, the method comprises: calculating the parameters of a model that approximates the noisy image as a function of the one or more guide channels (at the first resolution), and applying the calculated parameters to the one or more guide channels at the second resolution, to produce a denoised image at the second resolution.

Abstract: A method of rendering an image of a 3-D scene includes rendering a noisy image and obtaining one or more guide channels. For each of a plurality of local neighborhoods, the method comprises: calculating the parameters of a model that approximates the noisy image as a function of the one or more guide channels, and applying the calculated parameters to produce a denoised image. At least one of (i) the noisy image, (ii) the one or more guide channels, and (iii) the denoised image, are stored in a quantized low-bitdepth format.

Abstract: A method in a hardware implementation of a Convolutional Neural Network (CNN), includes receiving a first subset of data having at least a portion of weight data and at least a portion of input data for a CNN layer and performing, using at least one convolution engine, a convolution of the first subset of data to generate a first partial result; receiving a second subset of data comprising at least a portion of weight data and at least a portion of input data for the CNN layer and performing, using the at least one convolution engine, a convolution of the second subset of data to generate a second partial result; and combining the first partial result and the second partial result to generate at least a portion of convolved data for a layer of the CNN.

Abstract: A histogram-based method of selecting a fixed point number format for representing a set of values input to, or output from, a layer of a Deep Neural Network (DNN). The method comprises obtaining a histogram that represents an expected distribution of the set of values of the layer, each bin of the histogram is associated with a frequency value and a representative value in a floating point number format; quantising the representative values according to each of a plurality of potential fixed point number formats; estimating, for each of the plurality of potential fixed point number formats, the total quantisation error based on the frequency values of the histogram and a distance value for each bin that is based on the quantisation of the representative value for that bin; and selecting the fixed point number format associated with the smallest estimated total quantisation error as the optimum fixed point number format for representing the set of values of the layer.

Abstract: A method and data processing system for implementing a neural network containing at least one matrix multiplication operation. The matrix multiplication operation is mapped to a graph of neural network operations including at least one element-wise operation. The at least one element-wise operation is implemented in fixed-function hardware of a neural network accelerator.

Abstract: Methods and systems for performing a convolution transpose operation between an input tensor having a plurality of input elements and a filter comprising a plurality of filter weights. The method includes: dividing the filter into a plurality of sub-filters; performing, using hardware logic, a convolution operation between the input tensor and each of the plurality of sub-filters to generate a plurality of sub-output tensors, each sub-output tensor comprising a plurality of output elements; and interleaving, using hardware logic, the output elements of the plurality of sub-output tensors to form a final output tensor for the convolution transpose.