Abstract: A method is provided that includes receiving points of an image in which objects are depicted. The method includes performing a greedy nearest-neighbor (GNN) cluster analysis of the image to group the points of the image. The GNN cluster analysis includes grouping the points into a plurality of local GNN clusters, from a greedy analysis using a k-d tree in which the points are organized. The plurality of local GNN clusters are then extended into a plurality of global GNN clusters. This evaluating similarity of the local GNN clusters, merging the local GNN clusters into a global GNN cluster when each of the defined similarity criteria is evaluated to true, and passing the local GNN clusters as global GNN clusters when any of the defined similarity criteria is evaluated to false. The method then includes detecting the objects depicted in the image based on the global GNN clusters.

Abstract: The degree of differentiation of a cell in tissue is precisely determined. An estimating device (1) includes: a binarizing section (41) configured to generate binarized images from an image obtained by capturing an image of tissue; a Betti number calculating section (42) configured to calculate, for each binarized image, (i) the number of hole-shaped regions (b1) each surrounded by pixels of a first pixel value and each composed of pixels of a second pixel value, (ii) the number of connected regions each composed of the pixels of the first pixel value connected together, and (iii) a ratio (R) between (i) and (ii); a statistic calculating section (43) configured to calculate statistics of the calculated numbers (b1, b0) and ratio (R); and an estimating section (44) configured to feed input data including the calculated statistics to a trained estimating model to output the degree of differentiation of the cell in tissue.

Abstract: Systems and methods for providing assistance to a surgeon during an implant surgery are disclosed. A method includes defining areas of interest in diagnostic data of a patient and defining a screw bone type based on the surgeon's input. Post defining the areas of interest, salient points are determined for the areas of interest. Successively, an XZ angle, an XY angle, and a position entry point for a screw are determined based on the salient points of the areas of interest. Successively, a maximum screw diameter and a length of the screw are determined based on the salient points. Thereafter, the screw is identified and suggested to the surgeon for usage during the implant surgery.

Abstract: A vehicle exterior environment recognition apparatus includes first and second width derivation units, a degree-of-overlap derivation unit, and a specific object determination unit. The first width derivation unit derives a first width of a first three-dimensional object based on a luminance image. The second width derivation unit derives a second width of a second three-dimensional object based on a distance image. The degree-of-overlap derivation unit obtains an overlap width with which the first and second widths overlap horizontally, and set, as a degree of overlap, whichever is greater out of a value obtained by dividing the overlap width by the first width and a value obtained by dividing the overlap width by the second width. The specific object determination unit determines that the first and second three-dimensional objects are an identical specific object, on the condition that the degree of overlap is greater than a predetermined threshold.

Abstract: Embodiments of the present invention relate to a dual-light image integration method, a dual-light image integration device and an unmanned aerial vehicle (UAV). The method includes: receiving first image data from a first image device and second image data from a second image device; separately storing the first image data and the second image data; combining the first image data and the second image data to composite third image data; and transmitting the third image data. According to the present invention, dual-light images are processed by using different data processing methods depending on subsequent operations, which can maintain synchronization of the dual-light images during image transmission while avoiding information loss during image storage, thereby well satisfying user needs.

Abstract: In some examples, a device may receive, from a first camera, a plurality of images of an airspace corresponding to an area of operation of an unmanned aerial vehicle (UAV). The device may detect, based on the plurality of images from the first camera, a candidate object approaching or within the airspace. Based on detecting the candidate object, the device may control a second camera to direct a field of view of the second camera toward the candidate object. Further, based on images from the second camera captured at a first location and images from at least one other camera captured at a second location, the candidate object may be determined to be an object of interest. In addition, at least one action may be taken based on determining that the candidate object is the object of interest.

Abstract: Techniques related to implementing convolutional neural networks for face or other object recognition are discussed. Such techniques may include applying, in turn, a depth-wise separable convolution, a condense point-wise convolution, and an expansion point-wise convolution to input feature maps to generate output feature maps such that the output from the expansion point-wise convolution has more channels than the output from the condense point-wise convolution.

Abstract: A technique for assessing development condition of a property is provided that determines development condition for an individual property or properties of interest using image or other sensor data from one or more unmanned aerial vehicles taken over the development process. A property condition output may be generated to indicate a condition of the property or properties.

Abstract: Methods and apparatus for compressing image data are described along with corresponding methods and apparatus for decompressing the compressed image data. An encoder unit, which generates the compressed image data, comprises an input arranged to receive a first image and a second image, wherein the second image is twice the width and height of the first image, a prediction generator arranged to generate a prediction texture from the first image using an adaptive interpolator, a difference texture generator arranged to generate a difference texture from the prediction texture and the second image and in encoder unit arranged to encode the difference texture.

Abstract: The present invention has been made in view of the foregoing, and an object thereof is to provide an information processing device capable of appropriately determining a necessity of the ultrasonography even when a high-density mammary gland area exists in a narrow area. The present invention provides an information processing device comprising: a mammary gland area extractor configured to extract a mammary gland area in a mammography image; a mammary gland pixel detector configured to detect a mammary gland pixel in the mammary gland area; and a mammary gland density calculator configured to calculate a mammary gland density based on a ratio of the mammary gland pixels to the mammary gland area, wherein the mammary gland area is a narrower area than an entire breast in the mammography image.

Abstract: A method for detecting potential dents in a surface includes providing a flying drone with an image acquisition device and a light source, the shape of which is elongate, and a data processing device. At least one series of images is acquired of portions of the surface by the image acquisition device by moving the flying drone past the surface along a trajectory so that, for each image of the series, the light source is illuminating the corresponding portion of the surface. The data processing device is operated to analyze the shape of a specular reflection of the light source in images of the series of images to estimate the position of any dents in the surface.

Abstract: Provided is an excellent robot device capable of preferably detecting difference between dirt and a scratch on a lens of a camera and difference between dirt and a scratch on a hand. A robot device detects a site in which there is the dirt or the scratch using an image of the hand taken by a camera as a reference image. Further, this determines whether the detected dirt or scratch is due to the lens of the camera or the hand by moving the hand. The robot device performs cleaning work assuming that the dirt is detected, and then this detects the difference between the dirt and the scratch depending on whether the dirt is removed.

Abstract: This document describes techniques and systems for authentication management through IMU and radar. The techniques and systems use inertial sensor data from an inertial measurement unit (IMU) and/or radar data to manage authentication for a computing device. By so doing, the techniques conserve power, improve accuracy, or reduce latency relative to many common techniques and systems for computing-device authentication.

Abstract: A tracking method includes obtaining first characteristic information of a target object through a first camera; obtaining second characteristic information of the target object through a second camera; tracking the target object based on the first characteristic information through the first camera; obtaining attitude information of a first gimbal to which the first camera is mounted; and adjusting an attitude of a second gimbal to which the second camera is mounted based on the attitude information of the first gimbal to maintain consistency between the attitude of the second camera and an attitude of the first camera.

Abstract: The system and method are directed to a prototypical contrastive learning (PCL). The PCL explicitly encodes the hierarchical semantic structure of the dataset into the learned embedding space and prevents the network from exploiting low-level cues for solving the unsupervised learning task. The PCL includes prototypes as the latent variables to help find the maximum-likelihood estimation of the network parameters in an expectation-maximization framework. The PCL iteratively performs an E-step for finding prototypes with clustering and M-step for optimizing the network on a contrastive loss.

Abstract: Systems and methods for predicting images with enhanced spatial resolution using a neural network are provided herein. According to an aspect of the invention, a method includes accessing an input image of a biological sample, wherein the input image includes a first spatial resolution and a plurality of spectral images, and wherein each spectral image of the plurality of spectral images includes data from a different wavelength band at a different spectral channel; applying a trained artificial neural network to the input image; generating an output image at a second spatial resolution, wherein the second spatial resolution is higher than the first spatial resolution, and wherein the output image includes a fewer number of spectral channels than the plurality of spectral images included in the input image; and outputting the output image.

Abstract: A first robot arm places a calibration object into a load lock that separates a factory interface from a transfer chamber using a first taught position. A second robot arm retrieves the calibration object from the load lock using a second taught position. A controller determines, using a sensor, a first offset amount between a calibration object center of the calibration object and a pocket center of the second robot arm. The controller determines a characteristic error value that represents a misalignment between the first taught position of the first robot arm and the second taught position of the second robot arm based on the first offset amount. The first robot arm or the second robot arm uses the first characteristic error value to compensate for the misalignment for objects transferred between the first robot arm and the second robot arm via the load lock.

Abstract: One or more images are used to analyze permanent cropland. The images can be obtained from one or more sensors on an aerial vehicle including, but not limited to, UAVs. The analysis of the permanent cropland includes, but is not limited to, analyzing the plants growing in the permanent cropland or analyzing the permanent crops growing on the plants. In one embodiment, the analysis can include estimating a total yield of the permanent crops where the estimated permanent crop yield is more accurate than the estimated yield obtained using traditional methods.

Abstract: An example embodiment of the present invention provides a method of assessing the condition of a pavement site, comprising: (a) acquiring aerial images of the site from above, for example by an unmanned aerial system (UAS); (b) using photogrammetry tools to generate an orthomosaic that represents the airport pavement surface; (c) using image analysis tools and machine learning methods to determine the location and extent of defects in the pavement; (c) producing an image representation of the site and the defects, where the location and extent of defects are discernible from the image; (d) using software application techniques to store and present defect data and other related information for client-side user access.

Abstract: A device may receive sensor data and video data associated with a vehicle, and may process the sensor data, with a rule-based detector model, to determine whether a probability of a vehicle accident satisfies a first threshold. The device may preprocess acceleration data of the sensor data to generate calibrated acceleration data, and may process the calibrated acceleration data, with an anomaly detector model, to determine whether the calibrated acceleration data includes anomalies. The device may filter the sensor data to generate filtered sensor data, and may process the filtered sensor data and anomaly data, with a decision model, to determine whether the probability of the vehicle accident satisfies a second threshold. The device may process the filtered sensor data, the anomaly data, and the video data, with a machine learning model, to determine whether the vehicle accident has occurred, and may perform one or more actions.