A CNN operates on the disparity or motion outputs of a block matching hardware module, such as a DMPAC module, to produce refined disparity or motion streams which improve operations in images having ambiguous regions. As the block matching hardware module provides most of the processing, the CNN can be small and thus able to operate in real time, in contrast to CNNs which are performing all of the processing. In one example, the CNN operation is performed only if the block hardware module output confidence level is below a predetermined amount. The CNN can have a number of different configurations and still be sufficiently small to operate in real time on conventional platforms.

A method for intelligently measuring vehicle speed based on a binocular stereo vision system includes: training a Single Shot Multibox Detector neural network to obtain a license plate recognition model; calibrating the binocular stereo vision system to acquire parameters of two cameras; detecting the license plates in the captured video frames with the license plate recognition model, locating the license plate position; performing feature point extraction and stereo matching by a feature-based matching algorithm; screening and eliminating the matching point pairs, and reserving the coordinates of the matching point pair closest to the license plate center; performing stereo measurement on the screened matching point pair to get the spatial coordinates of the position; calculating and obtaining the speed of the target vehicle. The present invention is easy to install and adjust, could simultaneously recognize multiple trained features automatically, and better suit the intelligent transportation networks and IoT (Internet of Things).

A method for intelligently measuring vehicle speed based on a binocular stereo vision system includes: training a Single Shot Multibox Detector neural network to obtain a license plate recognition model; calibrating the binocular stereo vision system to acquire parameters of two cameras; detecting the license plates in the captured video frames with the license plate recognition model, locating the license plate position; performing feature point extraction and stereo matching by a feature-based matching algorithm; screening and eliminating the matching point pairs, and reserving the coordinates of the matching point pair closest to the license plate center; performing stereo measurement on the screened matching point pair to get the spatial coordinates of the position; calculating and obtaining the speed of the target vehicle. The present invention is easy to install and adjust, could simultaneously recognize multiple trained features automatically, and better suit the intelligent transportation networks and IoT (Internet of Things).

The purpose of the present invention is to provide a fall risk evaluation system whereby risk of falling of an elderly person or other person to be managed can be easily evaluated on the basis of a captured image of daily life, instead of by a physical therapist, etc. To achieve this purpose, the present invention is a fall risk evaluation system comprising a stereo camera and a fall risk evaluation device, the fall risk evaluation device being provided with: a person authentication unit for authenticating a person to be managed who has been imaged by the stereo camera; a person tracking unit for tracking the person to be managed who is authenticated by the person authentication unit; an action extraction unit for extracting walking by the person to be managed; a feature value calculation unit for calculating a feature value of the walking extracted by the action extraction unit; an integration unit for generating integrated data obtained by integrating the outputs of the person authentication unit, the person tracking unit, the action extraction unit, and the feature value calculation unit; a fall index calculation unit for calculating a fall index value of the person to be managed, on the basis of a plurality of integrated data generated by the integration unit; and a fall risk evaluation unit for comparing the fall index value calculated by the fall index calculation unit and a threshold value to evaluate the risk of falling of the person to be managed.

The purpose of the present invention is to provide a fall risk evaluation system whereby risk of falling of an elderly person or other person to be managed can be easily evaluated on the basis of a captured image of daily life, instead of by a physical therapist, etc. To achieve this purpose, the present invention is a fall risk evaluation system comprising a stereo camera and a fall risk evaluation device, the fall risk evaluation device being provided with: a person authentication unit for authenticating a person to be managed who has been imaged by the stereo camera; a person tracking unit for tracking the person to be managed who is authenticated by the person authentication unit; an action extraction unit for extracting walking by the person to be managed; a feature value calculation unit for calculating a feature value of the walking extracted by the action extraction unit; an integration unit for generating integrated data obtained by integrating the outputs of the person authentication unit, the person tracking unit, the action extraction unit, and the feature value calculation unit; a fall index calculation unit for calculating a fall index value of the person to be managed, on the basis of a plurality of integrated data generated by the integration unit; and a fall risk evaluation unit for comparing the fall index value calculated by the fall index calculation unit and a threshold value to evaluate the risk of falling of the person to be managed.

A learning method and a learning device for visual odometry based on an ORB feature of an image sequence are provided. The learning method includes: recording images, and constituting an original data set by means of the plurality of obtained images; performing ORB feature extraction on the images in the original data set to realize extraction of first key features; performing feature extraction and matching on continuous images in the original data set by means of a convolutional neural network, and extracting rich second key features from the sequential images; and inputting the first key features and the second key features extracted from the original data set into a multi-layer long-short-term memory network for training and learning, and generating and outputting estimation of a visual odometer. Rich first key features are extracted from an image sequence, and then a tracking algorithm is used for tracking the features in continuous frames.

A learning method and a learning device for visual odometry based on an ORB feature of an image sequence are provided. The learning method includes: recording images, and constituting an original data set by means of the plurality of obtained images; performing ORB feature extraction on the images in the original data set to realize extraction of first key features; performing feature extraction and matching on continuous images in the original data set by means of a convolutional neural network, and extracting rich second key features from the sequential images; and inputting the first key features and the second key features extracted from the original data set into a multi-layer long-short-term memory network for training and learning, and generating and outputting estimation of a visual odometer. Rich first key features are extracted from an image sequence, and then a tracking algorithm is used for tracking the features in continuous frames.

Disclosed herein is a system and method directed to object tracking and metric generation using a plurality of cameras. The system includes the plurality of cameras disposed around a playing surface in a mirrored configuration, where the plurality of cameras are time-synchronized. The system further includes logic that, when executed by a processor, causes performance of operations including: obtaining a sequence of images from the plurality of cameras, continuously detecting an object in image pairs at successive points in time, wherein each image pair corresponds to a single point in time, continuously determining a location of the object within the playing space through triangulation of the object within each image pair, detecting a player and the object within each image of a subset of image pairs of the sequence of images, identifying a sequence of interactions between the object and the player, and storing the sequence of interactions.

A six degrees of freedom (6 DoF) positioning tracking device and method, and an electronic apparatus are provided. The device includes at least three tracking cameras, an inertial navigation unit and a computation processing unit, where the at least three tracking cameras are distributed at preset angles and are configured to obtain image data in a current environment and transmit the image data to the computation processing unit; the inertial navigation unit is configured to obtain state information of the 6 DoF positioning tracking device in the current environment and transmit the state information to the computation processing unit; and the computation processing unit is configured to determine 6 DoF data of the 6 DoF positioning tracking device relative to the current environment according to the image data and the state information.

Apparatus and associated methods relate to a method of non-contact motion detection. A one-dimensional optical sensor detects motion of a target or objects on a conveyor belt through a continuous measurement of targets or objects and a real-time comparison of the pixel images captured by the one-dimensional optical sensor. In an illustrative embodiment, a one-dimensional sensor may be configured to determine motion of objects based on changes to the captured intensities of pixel images over time. The sensor may continually capture photoelectric pixel images and compare a current pixel image with a previous pixel image to determine a frame differential image value. The frame differential image value is evaluated against a predetermined threshold over a predetermined time period. Based on the evaluation, a signal is output indicating whether the objects on the conveyor belt are moving or jammed.