Abstract: This disclosure provides methods, devices, and systems for an artificial intelligence (AI) network. The present implementations more specifically relate to an AI network on an application specific integrated circuit (ASIC) operable as a reconfigurable multilayer image processor capable of implementing different AI models. In some aspects, each layer in the multilayer AI network includes a plurality of multiplier-accumulator (MAC) units, and at least one layer is partitioned into a plurality of blocks of MAC units that are reconfigurable to operate independently or to operate in one or more combinations of blocks of MAC units. The arrangement of the plurality of blocks of MAC units in the at least one layer enables implementation of one or more virtual layers, reconfiguration of the input depth size, reconfiguration of the output feature map size, or a combination thereof, which may be used to executes a desired AI model for image processing.

Abstract: Systems and methods are provided for determining a characteristic of video data. A set of N frames of the video data is obtained and filtered using at least one filter to produce a set of N×T blocks of filtered video data, where T is a partition size associated with the at least one filter. Each block in the set of N×T blocks is classified as either a first type block or a second type block. A subset of blocks in the set of N×T blocks is associated with a corresponding frame from the set of N frames. The characteristic of video data is determined based, at least in part, on the subset of blocks in the set of N×T blocks that are associated with the frame.

Abstract: Devices, systems, methods, and other embodiments associated with reducing digital image noise are described. In one embodiment, a method includes filtering a digital image with a plurality of adaptive filters, wherein the plurality of adaptive filters include a first filter configured to filter noise surrounding one or more edges in the digital image, and a second filter configured to filter noise caused by a block based encoding of the digital image. The method further includes reducing a compression artifact from selected pixels in the digital image, wherein the compression artifact is reduced by (i) combining an output from the first filter and an output from the second filter in response to the digital image being determined to be blocky, and (ii) not combining the output from the first filter with the output of the second filter in response to the digital image not being determined to be blocky.

Abstract: Systems and methods are provided for determining characteristics of video data. A frame of video data is obtained, where the frame is represented by pixel data. A value is assigned to an element of a detection array based on pixel data in a portion of the video frame corresponding to the element. A frequency transform of values of the detection array is determined, and a characteristic of the video data is extracted based on the output of the frequency transform.

Abstract: Devices, systems, methods, and other embodiments associated with reducing digital image noise are described. In one embodiment, a method includes filtering a digital image with a plurality of adaptive filters, wherein the plurality of adaptive filters include a first filter configured to filter noise surrounding one or more edges in the digital image, and a second filter configured to filter noise caused by a block based encoding of the digital image. The method further includes reducing a compression artifact from selected pixels in the digital image, wherein the compression artifact is reduced by (i) combining an output from the first filter and an output from the second filter in response to the digital image being determined to be blocky, and (ii) not combining the output from the first filter with the output of the second filter in response to the digital image not being determined to be blocky.

Abstract: Devices, systems, methods, and other embodiments associated with reducing digital image noise are described. In one embodiment, a method includes determining, on a per pixel basis, mosquito noise values associated with pixels of a digital image. The method determines, on a per pixel basis, block noise values associated with the digital image. The method filters the digital image with a plurality of adaptive filters. A compression artifact in the digital image is reduced. The compression artifact is reduced by combining filter outputs from the plurality of adaptive filters. The filter outputs are combined based, at least in part, on the mosquito noise values and the block noise values.

Abstract: Systems and methods are provided for determining characteristics of video data. A frame of video data is obtained, where the frame is represented by pixel data. A value is assigned to an element of a detection array based on pixel data in a portion of the video frame corresponding to the element. A frequency transform of values of the detection array is determined, and a characteristic of the video data is extracted based on the output of the frequency transform.

Abstract: Devices, systems, methods, and other embodiments associated with reducing digital image noise are described. In one embodiment, a method includes determining, on a per pixel basis, mosquito noise values associated with pixels of a digital image. The method determines, on a per pixel basis, block noise values associated with the digital image. The method filters the digital image with a plurality of adaptive filters. A compression artifact in the digital image is reduced. The compression artifact is reduced by combining filter outputs from the plurality of adaptive filters. The filter outputs are combined based, at least in part, on the mosquito noise values and the block noise values.