A non-transitory computer-readable recording medium stores therein an estimation program that causes a computer to execute a process including, identifying a first person who uses a first cart from a first image acquired by capturing inside a store, generating skeleton information of the first person, acquiring, by using the skeleton information, a first space in which the first person grasps a grip part of the first cart, and estimating a first scale of the first person based on the first space and length information of the first cart.

A three-dimensional gesture detection device and a three-dimensional gesture detection method are provided. The three-dimensional gesture detection device includes a node detection unit, a gesture recognition model and a gesture trajectory detection unit. The node detection unit obtains several nodes according to each of the hand frames of a continuous hand image. The gesture recognition model obtains confidence levels of several gesture categories. The gesture trajectory detection unit includes a weight analyzer, a gesture analyzer, a key point classifier and a trajectory analyzer. The weight analyzer obtains weights of the gesture categories through a user interface. The gesture analyzer performs a weighting calculation on the confidence levels of the gesture categories to analyze a gesture on each of the hand frames. The key point classifier classifies several key points from the nodes. The trajectory analyzer obtains an inertial trajectory of the gesture according to the key points.

Disclosed is a system for monitoring an activity performed by the subject. The system comprises a non-imaging sensor configured to detect the subject in a scan area, wherein the subject is detected by reflected waveform thereby. The system also comprises a processing arrangement communicably coupled to the non-imaging sensor, wherein the processing arrangement is configured to receive the reflected waveform from the non-imaging sensor, employ a first neural network to estimate the skeletal pose of the subject, feed a temporal succession of a plurality of skeletal poses of the subject to a second neural network, and determine the activity performed by the subject based on the temporal succession of the plurality of skeletal poses. Disclosed also is a method for monitoring an activity performed by the subject.

A treadmill according to the present invention includes: a frame; an endless belt; an endless-belt drive unit; a plurality of cameras provided on the frame such that the orientations and/or locations thereof can be adjusted; a motion-data acquisition unit that markerlessly acquires motion data of a subject by using image information obtained with camera images; a motion analysis unit that analyzes the motion of the subject by using the motion data; and a control unit that controls the endless-belt drive unit based on the motion data.

A region of space may be monitored for the presence or absence of one or more control objects, and object attributes and changes thereto may be interpreted as control information provided as input to a machine or application. In some embodiments, the region is monitored using a combination of scanning and image-based sensing.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating an optical flow label from a lidar point cloud. One of the methods includes obtaining data specifying a training example, including a first image of a scene in an environment captured at a first time point and a second image of the scene in the environment captured at a second time point. For each of a plurality of lidar points, a respective second corresponding pixel in the second image is obtained and a respective velocity estimate for the lidar point at the second time point is obtained. A respective first corresponding pixel in the first image is determined using the velocity estimate for the lidar point. A proxy optical flow ground truth for the training example is generated based on an estimate of optical flow of the pixel between the first and second images.

An information processing apparatus detects a person and a commodity product from image data. The information processing apparatus acquires, from the image data, a position of a skeleton of the person included in skeleton information on the detected person. The information processing apparatus specifies, based on the position of the skeleton of the person, a behavior of the person exhibiting with respect to the commodity product. The information processing apparatus specifies, based on the specified behavior of the person exhibiting with respect to the commodity product, a combination of an attribute of the commodity product and a degree of interest in the commodity product.

A speed estimation system includes: a detection module configured to determine bounding boxes of an object moving on a surface in images, respectively, captured using a camera; a solver module configured to, based on the bounding boxes, determine a homography of the surface by solving an optimization problem, where the solver module is not trained; and a speed module configured to, using the homography, determine a speed that the object is moving on the surface.

An apparatus for predicting whether an object moving across a surface will reach a target destination is provided, the apparatus comprising circuitry configured to: receive a first image and one or more subsequent second images from a camera; identify a location of an object on a surface in the first image; identify a location of the object on the surface in one or more of the second images; determine one or more motion characteristics of the object based on the location of the object in the first image and the location of the object in the one or more second images; generate a predicted path of the object across the surface based on a model of the surface and the motion characteristics of the object; and generate a prediction of whether the object will reach the target destination based on the predicted path of the object and the location of the target destination.

An electronic device and method for detection of intentional contact between an object and a body part of a player are provided. An electronic device receives a plurality of frames of a video from an imaging device. The electronic device tracks a movement of the object and a movement of the body part across the plurality of frames. The electronic device determines, based on the tracked movement, a frame of interest from the plurality of first frames in which degree of overlap between position coordinates of the object and position coordinates of the body part of the player exceeds an overlap threshold. The electronic device detects a contact between the object and the body part of the player based on the frame of interest. The electronic device applies a machine learning model on the frame of interest and the tracked movement to determine whether the contact is intentional.