Nippon Telegraph and Telephone Corporation

Dan Mikami

Mariko Isogawa

Kosuke Takahashi

Yoshinori Kusachi

Hideaki Kimata

Ayumi Matsumoto

Akira Kojima

Toshiaki Takeda

Hiroko Yabushita

A technique that enables, even if a video image corresponding to one period is not obtained, estimating a rotational state of an object is provided. The rotational state estimation device includes: an object image generation unitthat generates object images, each of the object images being an image of an object, from an input video image; and a rotational state estimation unitthat estimates a rotational state of the object by selecting, from among a plurality of hypotheses of the rotational state, a hypothesis of the rotational state, a likelihood of an image of the object resulting from the object in the object image at a certain time being rotated for tunit time based on the hypothesis of the rotational state being high, using the object image at a time t and the object image at a time t+twhere tis a predetermined integer of no less than 1.

Rotation state estimation device, method and program

A learning data generation device for generating learning data for learning a recognizer capable of estimating a contour of a sphere making spinning motion, with high accuracy, the sphere being recorded in a single camera video image, is provided. The learning data generation device includes: a spinning rate estimation unit that receives an input of a learning video image in which motion of a spinning sphere is recorded and an initial value of a size of a contour of the recorded sphere in the video image, sets a plurality of set values of the size of the contour based on the initial value, and obtains an estimated value of a spinning rate of the sphere based on the learning video image, for each of the set values; a contour determination unit that receives an input of a true value of the spinning rate of the sphere, the true value being obtained in advance for the learning video image, and determines at least any of a plurality of the set values respectively corresponding to a plurality of the estimated values selected in order of closeness to the true value, as a determined value of the contour; and a learning data output unit that outputs the learning video image and the determined value as learning data.

Learning data generation device, learning data generation method, and program

A height estimation method performed by a height estimation apparatus includes a first feature point extraction step of extracting a feature point coordinate, a first coordinate estimation step of estimating a coordinate of a first subject frame, a pre-generation step of deriving a height of the first subject frame and generating a distance addition pattern and a correction coefficient for an individual missing pattern, a second feature point extraction step of extracting a feature point coordinate from a second input image, a second coordinate estimation step of estimating a coordinate of a second subject frame and estimating a coordinate of an object frame, a subject data selection step of selecting the individual missing pattern and the correction coefficient in accordance with the feature point coordinate, an object data selection step of selecting an object height, and a height estimation step of adding up a distance between a feature point coordinate and another feature point coordinate extracted in accordance with the missing pattern and deriving an estimated value of a height of the subject in accordance with a result of adding up the distance, the correction coefficient, the object height, and the coordinates of the object frame.

Height estimation method, height estimation apparatus, and program

A technique for interpolating positional information of a target in an image of a frame in which positional information of the target has not been acquired. An information interpolation device includes: a target information acquisition unitacquiring target information that is information related to a target in an image of each frame composing an input video; an indicator determination unitdetermining, based on the target information, an indicator indicating validity as a starting frame for each frame, the starting frame starting predetermined image processing on a target; a starting frame determination unitdetermining a starting frame based on the determined indicator; and a target information interpolation unitinterpolating, when a frame in which positional information of a target is not included in target information exists among frames within a predetermined number from the determined starting frame, positional information of the target of the frame that does not include the positional information of the target by using positional information of the target included in target information of a frame other than the frame that does not include the positional information of the target.

Information completion apparatus, method and program

A virtual reality system is provided that includes a video presentation apparatus, the virtual reality system comprising: the video presentation apparatus which includes circuitry configured to: obtain as input a video sequence composed of a plurality of frames and mask information specifying a complementation target region in the video sequence; separate a frame into a foreground region and a background region based on binary images representing differences between the plurality of frames included in the video sequence; determine either one of patch-search-based completion and paste synthesis as a complementation method for the complementation target region based on a number of pixels belonging to the foreground region and located within a given distance from a periphery of the complementation target region; and complement the complementation target region in accordance with the complementation method; and a virtual reality head mounted display which presents the complemented video sequence to the user.

Virtual environment construction apparatus, video presentation apparatus, model learning apparatus, optimal depth decision apparatus, methods for the same, and program

Provided is technology for extracting a person region from an image, that can suppress preparation costs of learning data. Included are a composited learning data generating unit that generates, from already-existing learning data that is a set of an image including a person region and a mask indicating the person region, and a background image to serve as a background of a composited image, composited learning data that is a set of a composited image and a compositing mask indicating a person region in the composited image, and a learning unit that learns model parameters using the composited learning data. The composited learning data generating unit includes a compositing parameter generating unit that generates compositing parameters that are a set of an enlargement factor, a degree of translation, and a degree of rotation, using the mask of the learning data, and a composited image and compositing mask generating unit that extracts a compositing person region from an image in the learning data using a mask of the learning data, generates the composited image from the background image and the compositing person region, using the compositing parameters, generates the compositing mask from a mask generating image and the compositing person region that are the same size as the composited image, using the compositing parameters, and generates the composited learning data.

Region extraction model learning apparatus, region extraction model learning method, and program

A motion classification model learning apparatus that learns a model for early recognizing a motion is provided. A training data acquisition part acquiring training data configured with pairs of video information about a motion that can be classified into any of a plurality of categories according to characteristics of the motion and category information that is a correct label corresponding to the video information; a motion history image generation part generation a motion history image of the video information; and a model learning part learning a model that outputs a label that is the category information, with the motion history image as an input are included.

Motion classification model learning apparatus, motion classification apparatus, motion classification model learning method and program

A video and audio presentation device configured to present video and audio for sports training with less delay in physical reaction is provided. The device includes: an offset determination unit configured to determine a time-series offset tobtained by correcting a time difference ton a basis of a correction coefficient &#x3b1;, the time difference trepresenting a time difference between Aand A, or a time difference between Aand A, where a physical reaction Arepresents time-series data of physical reaction to an incoming object in a real environment, a physical reaction Arepresents time-series data of physical reaction to an incoming object presented in a virtual reality environment, and a physical reaction Arepresents time-series data of physical reaction to an incoming object presented in a semi-virtual reality environment; a video presentation unit configured to present video in the semi-virtual reality environment; and an audio presentation unit configured to shift audio corresponding to the video on a basis of the time-series offset tso that the audio precedes arrival of the video, and to present the shifted audio in the semi-virtual reality environment.

Video and audio presentation device, video and audio presentation method, and program

A video image editing device or the like, capable of editing a wraparound video image generated by using a plurality of video images captured by multi-viewpoint cameras to be comfortably viewable by viewers, is provided. Based on information about the positions and the sizes of X subjects, a polynomial expression regarding the position of the subject and a polynomial expression regarding the size of the subject are generated. Correction or interpolation of the positions of the H subjects is performed by a polynomial approximation curve using the polynomial expression regarding the position of the subject, correction or interpolation of the sizes of the N subjects is performed by a polynomial approximation curve using the polynomial expression regarding the size of the subject, and the sizes of the N subjects are expanded or contracted with an expansion/contraction parameter p. An intermediate image is generated from two images of the same size corresponding to the photographing devices adjacent to each other, and a wraparound video image is generated, with a parameter indicating the length of the wraparound video image being represented by Tp.

Video editing apparatus, method and program for the same

Preliminary experience of a match at a player's perspective is enabled. A virtual environment material storagehas stored therein virtual environment materials for reproducing a dynamic object and a static object on a virtual environment. A dynamic object sensing unitchronologically measures a position and a posture of the dynamic object in a real environment and generates position and posture information composed of one movement action. A presentation sequence acquisition unitobtains a presentation sequence including position and posture information of a plurality of different kinds of dynamic objects. A virtual environment construction unitsynthesizes a virtual environment material of the dynamic object and a virtual environment material of the static object based on the presentation sequence to construct the virtual environment.

Growing community of inventors

Yokosuka, Japan

Dan Mikami