Abstract: Dynamic data quantization may be applied to minimize the power consumption of a system that implements a convolutional neural network (CNN). Under such a quantization scheme, a quantized representation of a 3×3 array of m-bit activation values may include 9 n-bit mantissa values and one exponent shared between the n-bit mantissa values (n<m); and a quantized representation of a 3×3 kernel with p-bit parameter values may include 9 q-bit mantissa values and one exponent shared between the q-bit mantissa values (q<p). Convolution of the kernel with the activation data may include computing a dot product of the 9 n-bit mantissa values with the 9 q-bit mantissa values, and summing the shared exponents. In a CNN with multiple kernels, multiple computing units (each corresponding to one of the kernels) may receive the quantized representation of the 3×3 array of m-bit activation values from the same quantization-alignment module.

Abstract: Dynamic data quantization may be applied to minimize the power consumption of a system that implements a convolutional neural network (CNN). Under such a quantization scheme, a quantized representation of a 3×3 array of m-bit activation values may include 9 n-bit mantissa values and one exponent shared between the n-bit mantissa values (n<m); and a quantized representation of a 3×3 kernel with p-bit parameter values may include 9 q-bit mantissa values and one exponent shared between the q-bit mantissa values (q<p). Convolution of the kernel with the activation data may include computing a dot product of the 9 n-bit mantissa values with the 9 q-bit mantissa values, and summing the shared exponents. In a CNN with multiple kernels, multiple computing units (each corresponding to one of the kernels) may receive the quantized representation of the 3×3 array of m-bit activation values from the same quantization-alignment module.

Abstract: In a low power hardware architecture for handling accumulation overflows in a convolver unit, an accumulator of the convolver unit computes a running total by successively summing dot products from a dot product computation module during an accumulation cycle. In response to the running total overflowing the maximum or minimum value of a data storage element, the accumulator transmits an overflow indicator to a controller and sets its output equal to a positive or negative overflow value. In turn, the controller disables the dot product computation module by clock gating, clamping one of its inputs to zero and/or holding its inputs to constant values. At the end of the accumulation cycle, the output of the accumulator is sampled. In response to a clear signal being asserted, the dot product computation module is enabled, and the running total is set to zero for the start of the next accumulation cycle.

Abstract: Convolution with a 5×5 kernel involves computing the dot product of a 5×5 data block with a 5×5 kernel. Instead of computing this dot product as a single sum of 25 products, the dot product is computed as a sum of four partial sums, where each partial sum is computed as a dot product of a 3×3 data block with a 3×3 kernel. The four partial sums may be computed by a single 3×3 convolver unit over four time periods. During each time period, at least some of the weights received by the 3×3 convolver unit may correspond to a quadrant of weights from the 5×5 kernel. A shifter circuit provides shifted columns (left or right shifted) of the input data to the 3×3 convolver unit, allowing the 3×3 convolver unit access to the 3×3 data block that spatially corresponds to a particular quadrant of weights from the 5×5 kernel.

Abstract: A convolutional neural network is used to generate hash strings corresponding to object instances. The characteristic hash strings are used to recognize the same object instance depicted in images generated at different times and by different camera devices.

Abstract: Dynamic data quantization may be applied to minimize the power consumption of a system that implements a convolutional neural network (CNN). Under such a quantization scheme, a quantized representation of a 3×3 array of m-bit activation values may include 9 n-bit mantissa values and one exponent shared between the n-bit mantissa values (n<m); and a quantized representation of a 3×3 kernel with p-bit parameter values may include 9 q-bit mantissa values and one exponent shared between the q-bit mantissa values (q<p). Convolution of the kernel with the activation data may include computing a dot product of the 9 n-bit mantissa values with the 9 q-bit mantissa values, and summing the shared exponents. In a CNN with multiple kernels, multiple computing units (each corresponding to one of the kernels) may receive the quantized representation of the 3×3 array of m-bit activation values from the same quantization-alignment module.

Abstract: Labeled data is deterministically generated for training or validating machine learning models for image analysis. Approaches are described that allow this training data to be generated, for example, in real-time, and in response to the conditions at the location where images are generated by image sensors.

Abstract: A three-dimensional model of the environment of one or more camera devices is determined, in which image processing for inferring the model may be performed at the one or more camera devices.

Abstract: Systems, methods, and machine-readable media for determining a three-dimensional environment model of the environment of one or more camera devices, in which image processing for inferring the model may be performed at the camera devices, are described.

Abstract: Systems, methods, and machine-readable media for using a convolutional neural network to generate hash strings corresponding to object instances, and thereby use the characteristic hash strings to recognize the same object instance depicted in images generated at different times and by different camera devices.

Abstract: Systems, methods, and machine-readable media for deterministically generating labeled data for training or validating machine learning models for image analysis are described. Approaches described herein allow this training data to be generated, for example, in real-time, and in response to the conditions at the location where images are generated by image sensors.

Abstract: Systems, methods, and machine-readable media for deterministically generating labeled data for training or validating machine learning models for image analysis, and for using such machine learning models to determine the contents of real-domain images by using a domain transfer to synthetic-appearing images are described.