Nec Corporation

Nec C Orporation

Tetsuo Inoshita

Katsuhiko Takahashi

Yuichi Nakatani

Gaku Nakano

Asuka Ishii

Eiji Muramatsu

Kota Iwamoto

Soma Shiraishi

Jun Kobayashi

Hiroshi Yamada

Hiroyoshi Miyano

Hideo Yokoi

Michio Nagai

Takanori Inoue

Kazuki Tsuchimochi

In an object detection device, the plurality of object detection units output a score indicating a probability that a predetermined object exists for each partial region set with respect to inputted image data. The weight computation unit uses weight computation parameters to compute a weight for each of the plurality of object detection units on a basis of the image data and outputs of the plurality of object detection units, the weight being used when the scores outputted by the plurality of object detection units are merged. The merging unit merges the scores outputted by the plurality of object detection units for each partial region according to the weights computed by the weight computation unit. The first loss computation unit computes a difference between a ground truth label of the image data and the score merged by the merging unit as a first loss. Then, the first parameter correction unit corrects the weight computation parameters so as to reduce the first loss.

Object detection device, learned model generation method, and recording medium

The present disclosure provides an image processing apparatus capable of efficiently generating a learning model having a high accuracy. An image processing apparatus () includes a data acquisition unit (), a data generation unit (), a recognition accuracy calculation unit (), and a learning data output unit (). The data acquisition unit () acquires input image data. The data generation unit () converts the input image data by using a data conversion parameter and newly generates image data. The recognition accuracy calculation unit () calculates a recognition accuracy of the image data generated by the data generation unit () by using a learning model stored in advance. The learning data output unit () outputs, as learning data, the image data of which the recognition accuracy calculated by the recognition accuracy calculation unit () is lower than a first threshold.

Image conversion by using color distribution

In an object recognition device, an estimation means estimates a possession of the person based on a pose of a person in an image. An object detection means detects an object from surroundings of the person in the image. A weighting means sets weights with respect to an estimation result of the possession and a detection result of the object based on the image. A combination unit combines the estimation result of the possession and the detection result of the object by using the weights being set, and recognizes the possession of the person.

Object recognition device, object recognition method, and recording medium

The information processing device performs distillation learning of a student model using unknown data which a teacher model has not learned. The label distribution determination unit outputs an arbitrary label for the unknown data. The data generation unit outputs new generated data using an arbitrary label and unknown data as inputs. The distillation learning part performs distillation learning of the student model using the teacher model and using the generated data as an input.

Information processing device, learning method, and recording medium

In an object detection device, a plurality of object detection units output a score indicating probability that a predetermined object exists, for each partial region set to image data inputted. The weight computation unit computes weights for merging the scores outputted by the plurality of object detection units, using weight calculation parameters, based on the image data. The merging unit merges the scores outputted by the plurality of object detection units, for each partial region, with the weights computed by the weight computation unit. The target model object detection unit configured to output a score indicating probability that the predetermined object exists, for each partial region set to the image data. The first loss computation unit computes a first loss indicating a difference of the score of the target model object detection unit from a ground truth label of the image data and the score merged by the merging unit. The first parameter correction unit corrects parameters of the target model object detection unit to reduce the first loss.

Object detection device, learning method, and recording medium

To improve pose estimation accuracy, a pose estimation apparatus according to the present invention extracts a person area from an image, and generates person area image information, based on an image of the extracted person area. The pose estimation apparatus according to the present invention further extracts a joint point of a person from the image, and generates joint point information, based on the extracted joint point. Then, the pose estimation apparatus according to the present invention generates feature value information, based on both of the person area image information and the joint point information. Then, the pose estimation apparatus according to the present invention estimates a pose of a person included in the image, based on an estimation model in which the feature value information is an input, and pose estimation result is an output.

Pose estimation apparatus, learning apparatus, pose estimation method, and non-transitory computer-readable recording medium

In the object detection device, the plurality of object detection units output a score indicating a probability that a predetermined object exists, for each partial region set with respect to image data inputted. The weight computation unit computes a weight for each of the plurality of object detection units by using weight computation parameters and based on the image data. The weights are used when the scores outputted by the plurality of object detection units are merged. The weight redistribution unit changes the weight for a predetermined object detection unit, among the weights computed by the weight computation unit, to 0 and output the weights. The merging unit merges the scores outputted by the plurality of object detection units for each of the partial regions, by using the weights computed by the weight computation unit and including the weight changed by the weight redistribution unit. The loss computation unit computes a difference between a ground truth label of the image data and the merged score merged by the merging unit as a loss. Then, the parameter correction unit corrects the weight computation parameters so as to reduce the loss.

Image processing apparatus, image processing method, and non-transitory computer readable medium storing program

A selecting unit selects first to third moving image data from a plurality of frame images composing moving image data. A first generating unit generates first training data that is labeled data relating to a specific recognition target from the first moving image data. A first learning unit learns a first model recognizing the specific recognition target by using the first training data. A second generating unit generates second training data from the second moving image data by using the first model. A second learning unit learns a second model by using the second training data. A third generating unit generates third training data from the third moving image data by using the second model.

Training data generation apparatus

A selecting unit selects first moving image data and second moving image data from a plurality of frame images composing moving image data. A first generating unit generates first training data that is labeled data relating to a specific recognition target from the frame images composing the first moving image data. A learning unit learns a first model recognizing the specific recognition target by using the first training data. A second generating unit generates second training data that is labeled data relating to the specific recognition target from the frame images composing the second moving image data by using the first model.

Provided are a moving body guidance apparatus, a moving body guidance method and a computer-readable recording medium that are for accurately guiding a moving body to a target site. The moving body guidance apparatus has a detection unitthat detects, from an imagecaptured by an image capturing apparatusmounted on a moving body, a target member imagecaptured of an entirety of a target memberor a feature member imagecaptured of an entirety or a portion of feature membersandforming the target member, and a control unitthat performs guidance control for moving a set positionset with respect to the imageand indicating a position of the moving bodycloser to the target member imageor closer to a designated regionset based on the feature member image

Moving body guidance apparatus, moving body guidance method, and computer-readable recording medium

Provided is a target member for accurately guiding a moving body to a target site. A target memberis used when performing control for guiding a moving bodyto a target site, and the target memberis formed by using two or more different feature membersthat are set such that a shape of a target member imagethat corresponds to the target membercaptured by an image capturing unit(an image capturing apparatus) that is mounted in the moving bodychanges according to a measurement distance that indicates a distance between the target memberand the moving body

Target member

To more reliably land a flying body at a desired point, a flying body includes a determiner that determines whether the flying body is taking off and ascending from a takeoff point or descending to land, a camera mounted in the flying body, a recorder that records a lower image captured by the camera if it is determined that the flying body is taking off and ascending and a guider that, if it is determined that the flying body is descending to land, guides the flying body to the takeoff point while descending using a lower image recorded in the recorder during takeoff/ascent and a lower image captured during the descent.

Flying body control apparatus, flying body control method, and flying body control program

Provided is an object detection device for efficiently and simply selecting an image for creating instructor data on the basis of the number of detected objects. The object detection device is provided with: a detection unit for detecting an object from each of a plurality of input images using a dictionary; an acceptance unit for displaying, on a display device, a graph indicating the relationship between the input images and the number of subregions in which the objects are detected, and displaying, on the display device, in order to create instructor data, one input image among the plurality of input images in accordance with a position on the graph accepted by operation of an input device; a generation unit for generating the instructor data from the input image; and a learning unit for learning a dictionary from the instructor data.

Object detection device, object detection method, and recording medium

Concerning a partial area image that constitutes a wide area image, to control a flying body in accordance with a flight altitude at a past point of time of image capturing, an information processing apparatus includes a wide area image generator that extracts, from a flying body video obtained when a flying body captures a ground area spreading below while moving, a plurality of video frame images and combines the video frame images, thereby generating a captured image in a wide area, an image capturing altitude acquirer that acquires a flight altitude at a point of time of image capturing by the flying body for each of the plurality of video frame images, and an image capturing altitude output unit that outputs a difference of the flight altitude for each video frame image.

Growing community of inventors

Tokyo, Japan

Tetsuo Inoshita