Abstract: An authentication system includes: a first imaging device that images an individual object in a first stage to generate a registration image; a second imaging device that images an individual object to be authenticated in the second stage to generate an authentication image; and a registration device configured to authenticate the individual object to be authenticated, by extracting a registration partial image from the registration image with use of a registration template to acquire registration image information, extracting an authentication partial image from the authentication image with use of an authentication template to acquire authentication image information, and determining whether the authentication image information and the registration image information represent a same individual object, the authentication template corresponding to the registration template, and having, with respect to the registration template, a difference in accordance with a treatment/processing.

Abstract: A neural network apparatus (20) includes a storage unit (24) storing a neural network model, and an arithmetic unit (22) inputting input information into an input layer of the neural network and outputting an output layer. A weight matrix (W) of an FC layer of the neural network model is constituted by a product of a weight basis matrix (Mw) of integers and a weight coefficient matrix (Cw) of real numbers. In the FC layer, the arithmetic unit (22) uses an output vector from a previous layer as an input vector (x) to decompose the input vector (x) into a product of a binary input basis matrix (Mx) and an input coefficient vector (cx) of real numbers and an input bias (bx) and derives a product of the input vector (x) and a weight matrix (W).

Abstract: An individual object identification system includes: an image acquisition processor configured to perform an image acquisition process to acquire an image of a subject acquired using imaging equipment; a feature point extraction processor that extracts a feature point from the image; a local feature amount calculation processor that calculates a local feature amount of the feature point; a local feature amount group classification processor that performs classification into a predetermined number of local feature amount groups; a global feature amount calculation processor that calculates a global feature amount based on each of the local feature amount groups; a searching target image registration processor that registers a plurality of images that are searching targets; a global feature amount registration processor that registers the global feature amount related to each of the registered images in a global feature amount registration unit; a narrowing processor that narrows down the plurality of registere

Abstract: In a data processing apparatus, an M×M data processing unit performs M×M convolution processing using data from an input buffer unit. An N×N data processing unit performs N×N convolution processing using the data from the input buffer unit. A first output buffer unit stores one of results of processing by the M×M data processing unit and the N×N data processing unit, and outputs the same to the input buffer unit. A second output buffer unit stores the other of the results of processing by the M×M data processing unit and the N×N data processing unit. The second output buffer unit transfers the result of processing to the external memory.

Abstract: A feature point position estimation device is provided. The feature point position estimation device includes a subject detection section for detecting a subject region from a subject image, a feature point positioning section for positioning a feature point at a preliminarily prepared initial feature point position with respect to the subject region, a feature amount acquisition unit for acquiring a feature amount of the feature points arranged, a regression calculation unit for calculating a deviation amount of a position of a true feature point with respect to the position of the feature point by performing a regression calculation on the feature amount, and a repositioning unit for repositioning the feature points based on the deviation amount. The regression calculation unit calculates the deviation amount by converting the feature amount in a matrix-resolved regression matrix.

Abstract: By an information process device, a data decomposition method, or a data decomposition program stored in a computer-readable non-transitory storage medium, model data including multiple values is approximated by approximate data including a combination of basis data and coefficient data. A basis data candidate that constitutes the approximate data is selected. An approximate data candidate and an evaluation metric that evaluates the approximate data candidate are calculated. A regression model representing a relationship between the evaluation metric and the basis data candidate is generated. The selection, calculation, and generation are executed at least once. The coefficient data is calculated. The basis data candidate is selected to cause the regression model to more accurately predict the evaluation metric.

Abstract: A feature point position estimation device is provided. The feature point position estimation device includes a subject detection section for detecting a subject region from a subject image, a feature point positioning section for positioning a feature point at a preliminarily prepared initial feature point position with respect to the subject region, a feature amount acquisition unit for acquiring a feature amount of the feature points arranged, a regression calculation unit for calculating a deviation amount of a position of a true feature point with respect to the position of the feature point by performing a regression calculation on the feature amount, and a repositioning unit for repositioning the feature points based on the deviation amount. The regression calculation unit calculates the deviation amount by converting the feature amount in a matrix-resolved regression matrix.

Abstract: A recognition system includes: a sensor processing unit (SPU) that performs sensing to output a sensor value; a task-specific unit (TSU) including an object detection part that performs an object detection task based on the sensor value and a semantic segmentation part that performs a semantic segmentation task based on the sensor value; and a generic-feature extraction part (GEU) including a generic neural network disposed between the sensor processing unit and the task-specific unit, the generic neural network being configured to receive the sensor value as an input to extract a generic feature to be input in common into the object detection part and the semantic segmentation part.

Abstract: A neural network apparatus (20) includes a storage unit (24) storing a neural network model, and an arithmetic unit (22) inputting input information into an input layer of the neural network and outputting an output layer. A weight matrix (W) of an FC layer of the neural network model is constituted by a product of a weight basis matrix (Mw) of integers and a weight coefficient matrix (Cw) of real numbers. In the FC layer, the arithmetic unit (22) uses an output vector from a previous layer as an input vector (x) to decompose the input vector (x) into a product of a binary input basis matrix (Mx) and an input coefficient vector (cx) of real numbers and an input bias (bx) and derives a product of the input vector (x) and a weight matrix (W).

Abstract: A relatedness determination device includes: a feature vector acquisition portion that acquires a binarized feature vector; a basis vector acquisition portion that acquires a plurality of basis vectors obtained by decomposing a real vector into a linear sum of the basis vectors, which have a plurality of elements including only binary or ternary discrete values; and a vector operation portion that sequentially performs inner product calculation between the binarized feature vector and each of the basis vectors to determine relatedness between the real vector and the binarized feature vector.

Abstract: A recognition system includes: a sensor processing unit (SPU) that performs sensing to output a sensor value; a task-specific unit (TSU) including an object detection part that performs an object detection task based on the sensor value and a semantic segmentation part that performs a semantic segmentation task based on the sensor value; and a generic-feature extraction part (GEU) including a generic neural network disposed between the sensor processing unit and the task-specific unit, the generic neural network being configured to receive the sensor value as an input to extract a generic feature to be input in common into the object detection part and the semantic segmentation part.

Abstract: A feature amount conversion apparatus includes a plurality of bit rearrangement units, a plurality of logical operation units, and a feature integration unit. The bit rearrangement units generate rearranged bit strings by rearranging elements of an inputted binary feature vector into diverse arrangements. The logical operation units generate logically-operated bit strings by performing a logical operation on the inputted feature vector and each of the rearranged bit strings. The feature integration unit generates a nonlinearly converted feature vector by integrating the generated logically-operated bit strings.

Abstract: A relatedness determination device includes: a feature vector acquisition portion that acquires a binarized feature vector; a basis vector acquisition portion that acquires a plurality of basis vectors obtained by decomposing a real vector into a linear sum of the basis vectors, which have a plurality of elements including only binary or ternary discrete values; and a vector operation portion that sequentially performs inner product calculation between the binarized feature vector and each of the basis vectors to determine relatedness between the real vector and the binarized feature vector.

Abstract: A bit code converter transforms a learning feature vector using a transformation matrix updated by a transformation matrix update unit, and converts the transformed learning feature vector into a bit code. When the transformation matrix update unit substitutes a substitution candidate for an element of the transformation matrix, a cost function calculator fixes the substitution candidate that minimizes a cost function as the element. The transformation matrix update unit selects the element while sequentially changing the elements, and the cost function calculator fixes the selected element every time the transformation matrix update unit selects the element, thereby finally fixing the optimum transformation matrix.