Abstract: One example method includes identifying dissimilar items in a data set. A data set may be walked one or more times and the nodes or vertices of the data set may be scored based on the number of times the nodes are touched during the walks. Scores below a threshold score are determined to be dissimilar nodes in the data set. This allows a diverse set of nodes to be identified. A dissimilar data set may be used to prevent unintentional bias in algorithmic training.

Abstract: An object detection and classification verification system for a vehicle includes a projection system configured to project a three-dimensional (3D) scene pre-captured at a known distance and comprising at least one known object onto a surface in front of the vehicle a controller configured to verify a performance of an object detection and classification routine by performing the object detection and classification routine on the projected 3D scene to generate a set of results, comparing the set of results to a set of expected results associated with the projected 3D scene, and based on the comparing, determining whether the performance of the object detection and classification routine satisfies a predetermined threshold metric.

Abstract: Systems and methods for automatic detection of surgical specialty type and procedure type are disclosed. One or more classification networks may be applied to automatically process input surgical image data in order to recognize and determine a surgical specialty type and a surgical procedure type depicted in the input image data. Based on the determination made by the system, one or more output indications may be generated and one or more surgical devices may be automatically controlled, such as by being optimized for use during the surgical procedure type and/or surgical specialty type represented by the input image.

Abstract: There is provided a learning method and a learning device for performing transfer learning on an object detector that has been trained to detect first object classes such that the object detector is able to detect second object classes. Further, a testing method and a testing device are provided to allow at least part of the first object classes and the second object classes to be detected by using the object detector having been trained through the transfer learning. Accordingly, a detection performance can be improved for the second object classes that cannot be detected through training data set corresponding to the first object classes.

Abstract: This application provides a method of generating a description for a picture set performed at a computer device, and a storage medium. The method includes: acquiring a picture set to be processed; performing picture feature extraction on each picture in the picture set to acquire a picture feature corresponding to the picture, and forming a picture feature sequence corresponding to the picture set by using the picture features corresponding to the pictures; performing scene feature extraction on a picture feature sub-sequence corresponding to each scene in the picture feature sequence to acquire a scene feature corresponding to the scene, and forming a scene feature sequence corresponding to the picture set by using the scene features corresponding to the scenes; and generating textual description information of the picture set according to the picture feature sequence and the scene feature sequence that correspond to the picture set.

Abstract: Methods, systems, and computer-readable storage media for determining that a subject is a live person include capturing one or more images of two eyes of a subject. The one or more images, from each of the two eyes are used to obtain respective corneal reflections. Depth information associated with a scene in front of the subject is determined, based on an offset between the respective corneal reflections. A determination is made, based at least on the depth information, that the subject is a live person. Responsive to determining that the subject is a live person, an authentication process is initiated to authenticate the subject.

Abstract: Systems and methods are disclosed to objectively identify sub-second behavioral modules in the three-dimensional (3D) video data that represents the motion of a subject. Defining behavioral modules based upon structure in the 3D video data itself— rather than using a priori definitions for what should constitute a measurable unit of action— identifies a previously-unexplored sub-second regularity that defines a timescale upon which behavior is organized, yields important information about the components and structure of behavior, offers insight into the nature of behavioral change in the subject, and enables objective discovery of subtle alterations in patterned action. The systems and methods of the invention can be applied to drug or gene therapy classification, drug or gene therapy screening, disease study including early detection of the onset of a disease, toxicology research, side-effect study, learning and memory process study, anxiety study, and analysis in consumer behavior.

Abstract: A device includes: a distribution information acquiring part configured to acquire, based on an image in which a plurality of cells that are cultivated in a predetermined area are imaged, distribution information relating to a distribution in the predetermined area of the plurality of cells; and a determination part configured to determine a cultivated state of the plurality of cells based on the distribution information acquired by the distribution information acquiring part.

Abstract: A system and method for providing image guidance for placement of one or more medical devices at a target location. The system can determine one or more intersections between a medical device and an image region based at least in part on first emplacement data and second emplacement data. Using the determined intersections, the system can cause one or more displays to display perspective views of image guidance cues, including an intersection indicator in a virtual 3D space.

Abstract: An example method includes capturing, via an x-ray machine, a plurality of x-ray images of a patient covering a number of different anatomy of the patient in any order, using a machine learning algorithm to process the plurality of x-ray images for identification of an anatomy in respective x-ray images of the plurality of x-ray images, associating a label with each of the plurality of x-ray images based on the identification of the anatomy, positioning each of the plurality of x-ray images upright based on a preset coordinate scheme for the anatomy, arranging the plurality of x-ray images into a predetermined order based on the species of the patient, and generating and outputting a data file including the plurality of x-ray images in the predetermined order, positioned upright, and labeled.

Abstract: Methods and systems are provided for medical imaging systems. In one embodiment, a method for a medical imaging system comprises acquiring emission data during a positron emission tomography (PET) scan of a patient, reconstructing a series of live PET images while acquiring the emission data, tracking motion of the patient during the acquiring by determining a per-voxel variation for selected voxels in a current live PET image of the series of live PET images, and outputting an indication of patient motion based on the per-voxel variation for the selected voxels in each live PET image. In this way, patient motion during the scan may be identified and compensated for via scan acquisition and/or data processing adjustments, thereby producing a diagnostic PET image with reduced motion artifacts and increased diagnostic quality.

Abstract: The invention relates to a method for creating a semantic representation of the environment of a vehicle, having the following steps: acquiring (S1) sensor measuring data by means of at least one first (2a) and/or second environment detection sensor (2b) of the vehicle, detecting (S2) objects and free spaces in the environment of the vehicle, creating (S3) a grid map having occupied and free grid cells based on the detected objects and free spaces; determining (S4) height information by means of the at least one first (2a) and/or second environment detection sensor (2b); adding (S5) the height information to each grid cell; allocating (S6) semantic information to the grid cells based on the height information.

Abstract: An inspection apparatus (100) detects an inspection object (90) from first image data (10) in which the inspection object (90) is included. The inspection apparatus (100) generates second image data (20) by performing a geometric transform on the first image data (10) in such a way that a view of the detected inspection object (90) becomes a view satisfying a predetermined reference. In an inference phase, the inspection apparatus (100) detects, by using an identification model for detecting an abnormality of the inspection object (90), an abnormality of the inspection object (90) included in the second image data (20). Further, in a learning phase, the inspection apparatus (100) learns, by using the second image data (20), an identification model for detecting an abnormality of the inspection object (90).

Abstract: The present disclosure provides a method and device to process identity information based on a fundus image. An identity information comparison method includes: recognizing a fundus image by using a neural network to obtain a multi-dimensional feature vector representing the identity of a user; comparing the obtained multi-dimensional feature vector with each multi-dimensional feature vector stored in a database; and determining, according to a comparison result, whether there is a matching between the currently obtained multi-dimensional feature vector and a multi-dimensional feature vector pre-stored in the database.

Abstract: The present teaching relates to method, system, and programming for responding to an image related query. Information related to each of a plurality of images is received, wherein the information represents concepts co-existing in the image. Visual semantics for each of the plurality of images are created based on the information related thereto. Representations of scenes of the plurality of images are obtained via machine learning, based on the visual semantics of the plurality of images, wherein the representations capture concepts associated with the scenes.

Abstract: A system and method for deformation simulation of a hollow organ able to be deformed by the introduction of a medical instrument. The method includes provision of a pre-trained machine-learning algorithm, provision of a 3D medical recording of the hollow organ with surrounding tissue, with the 3D recording having been recorded before the introduction of a medical instrument, segmentation or provision of a segmentation of the 3D medical recording of the hollow organ and establishment or provision of a three-dimensional model of the hollow organ, provision of information about a medical instrument introduced or to be introduced, and simulation of the deformation of the hollow organ to be expected from introduction of the instrument on the basis of the segmented 3D medical recording of the hollow organ and of the surrounding tissue and of the information about the instrument by using the pre-trained machine-learning algorithm.

Abstract: Some implementations involve a process that identifies a subset of points in an image and creates descriptors for these points. The detection and descriptor process may use one or more neural networks. In some implementations, the process includes a neural network that uses one or more fixed (e.g., weight independent) neural network layers to perform certain functions that can be performed more accurately and/or efficiently than via non-fixed (e.g., weight-based) layers. In some implementations, for example, a neural network includes a layer that determines orientation formulaically within the neural network. Such orientations may be determined convolutionally (e.g., using sliding patches) but are not determined based on internal node weights within the layer that were determined during the training of the neural network.

Abstract: Two sets of data, each containing property listings, are obtained from two discrete merchant platforms. Each property listing in a set of data of a first merchant is sequentially paired with each of the property listings in a set of data of a second merchant. For each pair, each image of the property listing of the first merchant is compared to each image of the property listing of the second merchant, and images of statistically sufficient similarity are identified. The similarity of images, and in particular, of similar images likely to be rooms of the property, are considered in a determination of whether the product listings of the first and second merchant are for the same cross-listed product.

Abstract: An image recognition device and method retrieve information on a marker. The marker encodes information identifying an asset by encoding the information in a binary pattern as recesses and non-recesses in the marker. The image recognition is performed by imaging the marker, mapping contrast in the image, identifying variations in the contrast, creating a mesh overlaying the image, identifying the present or absence of recesses from the mesh, and reading the binary pattern represented by the recesses in the marker.

Abstract: Methods and systems for determining rotation and clipping parameters for images of unit load devices (ULDs) are disclosed herein. An example method includes capturing a set of image data featuring a ULD. The example method may further include locating a fiducial marker proximate to the ULD within the set of image data. The example method may further include cropping the set of image data, based upon the located fiducial marker, to generate a set of marker point data and a set of floor point data. The example method may further include rotating the set of image data based upon the set of marker point data and the set of floor point data, and clipping the rotated set of image data based upon the set of marker point data and the set of floor point data.