Abstract: The present disclosure generally pertains to an object detection circuitry configured to: obtain first feature data which are based on first sensing data of a first sensor; compare the first feature data to a first predetermined feature model being representative of a predefined object, wherein the first predetermined feature model is specific for the first sensor, thereby generating first object probability data; obtain second feature data which are based on second sensing data of a second sensor; compare the second feature data to a second predetermined feature model being representative of the predefined object, wherein the second predetermined feature model is specific for the second sensor, thereby generating second object probability data; and combine the first and the second object probability data, thereby generating combined probability data for detecting the predefined object.

Abstract: A method for establishing a non-rigid multi-modal medical image registration model includes establishing a generative adversarial network GAN_dr; performing calculation with respect to structural representations of a reference image, a floating image, and an actual registered image in each sample in a medical dataset; using a calculation result to train GAN_dr; establishing a generative adversarial network GAN_ie; using the trained G_dr to generate the deformation recovered structural representations corresponding to each sample in the medical dataset; training GAN_ie; and after connecting the trained G_ie to G_dr, obtaining the registration model.

Abstract: In some implementations, a computing device in communication with an immersive content generation system may generate a first set of user interface elements configured to receive a first selection of a shape of a virtual stage light. In addition, the device may generate a second set of user interface elements configured to receive a second selection of an image for the virtual stage light. Also, the device may generate a third set of user interface elements configured to receive a third selection of a position and an orientation of the virtual stage light. Further, the generate a fourth set of user interface elements configured to receive a fourth selection of a color for the virtual stage light. Numerous other aspects are described.

Abstract: A computer device obtains a confocal microscopy image. The device determines a pinhole diameter of a first detector pinhole of a confocal microscope that is used to acquire the image. The pinhole diameter has an influence on a parameter of the image. The device obtains a target image processing model corresponding to the pinhole diameter of the first detector pinhole. The target image processing model is configured to improve the parameter of the image. The device causes the target image processing model to process the image to obtain a target image having the improved parameter.

Abstract: A method and apparatus for locating head side acupuncture points, a moxibustion robot, and a storage medium are provided. The method includes: training an acupuncture point identification model based on a head side acupuncture point sample set to obtain a target acupuncture point identification model, where the head side acupuncture point sample set includes a plurality of acupuncture point sample points and a plurality of support sample points in a curve of gallbladder meridian of foot-shaoyang; inputting a head side image to be identified to the target acupuncture point identification model to obtain a plurality of acupuncture points and a plurality of support points; fitting the plurality of acupuncture points and the plurality of support points to obtain a fitted curve; and determining whether the acupuncture points require correction based on the fitted curve, if yes, correcting the acupuncture points based on the fitted curve to obtain target acupuncture points.

Abstract: A method, apparatus and system for efficient navigation in a navigation space includes determining semantic features and respective 3D positional information of the semantic features for scenes of captured image content and depth-related content in the navigation space, combining information of the determined semantic features of the scene with respective 3D positional information using neural networks to determine an intermediate representation of the scene which provides information regarding positions of the semantic features in the scene and spatial relationships among the semantic features, and using the information regarding the positions of the semantic features and the spatial relationships among the semantic features in a machine learning process to provide at least one of a navigation path in the navigation space, a model of the navigation space, and an explanation of a navigation action by the single, mobile agent in the navigation space.

Abstract: In non-limiting examples of the present disclosure, systems, methods and devices for generating searchable data structures using observational context data are provided. An indication to process observational context data surfaced by a local computing device may be received. The observational context data may comprise display data or audio data. The observational context data may be processed by a context analysis engine. A determination may be made of a current context type occurring on the local computing device. The current context type may be determined from a plurality of context types, each having a semantic data structure comprising a unique set of entity types. A canvas comprising a searchable data structure of the current context type may be generated and populated with objects having entity types corresponding to the current context type. A displayable representation of the canvas may be displayed.

Abstract: A golf ball tracking system identifies a first pixel at which a golf ball is located in a first image frame and a second pixel at which the golf ball is located in a second image frame. The system may determine that the first pixel represents a location where the golf ball impacts the golf surface based at least in part on a difference in location between the first pixel in the first image frame and the second pixel in the second image frame.

Abstract: A system for updating continuous image alignment of separate cameras identifies a previous alignment matrix associated with a previous frame pair captured at one or more previous timepoints by a reference camera and a match camera. The previous alignment matrix is based on visual correspondences in the previous frame pair. The system also identifies a current matrix associated with a current frame pair captured at one or more current timepoints by the reference camera and the match camera. The current matrix is based on visual correspondences in the current frame pair. The system also identifies a difference value associated with the reference camera or the match camera relative to the one or more previous timepoints and the one or more current timepoints. The system also generates an updated alignment matrix by using the previous alignment matrix, the current matrix, and the difference value as inputs.

Abstract: Disclosed is a method of high-precision 3D reconstruction of existing railway track lines based on UAV multi-view images, including: acquiring initial data, acquiring a UAV image rail top centerline, calculating a rail top centerline based on a nonlinear least squares method, and calculating three-dimensional coordinates of the rail centerline. Based on the multi-view geometry principle in computer vision and photogrammetry, object space coordinates of the line can be directly calculated by using image information, which does not require outdoor workers to work online and can effectively improve the safety of railway operation line surveying and mapping. Therefore, this method has important engineering application value and application prospect.

Abstract: A line-scan camera calibration method based on an auxiliary camera, including: calibrating the auxiliary camera, rigidly combining the auxiliary camera with a line-scan camera and installing the combined camera onto a motion platform, and placing a checkerboard calibration board within an imaging area of the combined camera; driving a motion platform to capture images and dynamic scanning images of the checkerboard calibration board, calculating a motion vector of the combined camera; calculating coordinates of registration points; fitting an equation of an imaging plane using coordinates of multiple registration points in the auxiliary camera coordinate system; calculating internal parameters of the line-scan camera and a rigid transformation matrix between the auxiliary camera and the line-scan camera based on properties of a rotation matrix and the line-scan camera imaging model. This application improves the calibration accuracy of the line-scan camera.

Abstract: System, methods, and other embodiments described herein relate to an improved approach to training a depth model to derive depth estimates from monocular images using histograms to assess photometric losses. In one embodiment, a method includes determining loss values according to a photometric loss function. The loss values are associated with a depth map derived from an input image that is a monocular image. The method includes generating histograms for the loss values corresponding to different regions of a target image. The method includes, responsive to identifying erroneous values of the loss values, masking the erroneous values to avoid considering the erroneous values during training of the depth model.

Abstract: A vehicle external environment recognition apparatus includes at least one processor, and at least one memory coupled to the at least one processor. The at least one processor is configured to operate in cooperation with at least one program stored in the at least one memory to execute processing. The processing includes generating a distance image from luminance images, specifying, by using semantic segmentation, a floating matter class in the luminance images, and invalidating parallax associated with floating pixels that are included in the distance image and belong to the floating matter class.

Abstract: An online document system can allow users to upload a document packages containing multiple individual document files. The online document system then automatically identifies and separates the component documents within an uploaded document package based on a package template for that type of document package. Based on the documents included in the document package, the online document system can automatically take action on the identified documents. The online document system may recognize component documents within a document package using a set of rules defining potential component documents that may appear within the received document package and one or more methods of recognizing each type of component document. In some embodiments, the online document system can improve package templates and recognition of component documents over time using feedback from the importing users through updates to the document identification rules and/or a supplementary probabilistic recognition system.

Abstract: The present disclosure relates to the field of ultrasonic detection, in particular to a craniocerebral ultrasonic standard plane imaging and automatic detection and display method for abnormal regions. The following technical solution is adopted: contour detection is performed on the scanned ultrasound images the skull to construct a skull surface model, and the standard planes in ultrasound images are identified and extracted according to the skull surface model. The symmetry of the standard planes in ultrasound images is used to compare the similarity, so as to obtain the abnormal regions for segmentation and display. The advantages of the present disclosure are as follows: the skull surface model is constructed by detecting the cranial edge, and the coordinate system is established based on this model, thereby the standard planes in ultrasound images can be quickly identified from the scanned ultrasound images.

Abstract: An anesthetic equipment storage and waste air management module configured to housing electronic and electromechanical surgical equipment including a system to measure and record administration of one or more IV medications or fluids for IV administration. The module can include a housing having a lower section and a tower-like upper section, wherein the lower section is configured to house unrelated waste heat-producing electronic and electromechanical surgical equipment. The module can also include a cowling that substantially confines waste heat generated by the unrelated waste heat-producing electronic and electromechanical surgical equipment, and can include a system for measuring and recording the administration of the one or more IV medications and fluids.

Abstract: This application provides an image processing method, including: adaptively determining at least one interpolation point based on feature information or dynamic metadata of a to-be-processed image, and determining the at least one interpolation point and a modified spline curve that passes through the at least one interpolation point, where the modified spline curve includes at least one of a linear spline curve or a cubic spline curve; and performing tone mapping on the to-be-processed image based on a modified tone mapping curve, where the modified tone mapping curve includes at least a part of an initial tone mapping curve and the modified spline curve. According to the method provided in this application, the tone mapping curve can be adaptively modified based on a scene requirement, so that images in various scenes can have a good display effect.

Abstract: The present disclosure relates to a method, a device and a medium for performing vessel segmentation in a medical image. The method may comprise acquiring a medical image for vessel segmentation containing multiple parts, each of which contains vessels with different structural attributes. The method may comprise dividing the medical image into sub-medical images according to the parts by using a processor. The method may comprise determining individual vessel segmentation result for each part by means of using the vessel segmentation model corresponding to the part based on the sub-medical image of the part by using the processor. The method may comprise obtaining a vessel segmentation result of the medical image by means of fusing the individual vessel segmentation results of the sub-medical images of the parts by the processor.

Abstract: Disclosed herein are an apparatus and method for image outpainting based on learning of a single input image. The apparatus generates multiple extended images from an input image using multiple noise vectors, inputs a first extended image, among the multiple extended images, the input image, and an image acquired by horizontally flipping the input image and thereby trains a discriminator network such that the loss of a generative adversarial network is minimized, and learns the multiple extended images such that the sum of two differences is minimized, the first one being the difference between the input image and the center region of a second extended image, and the second one being the difference between images acquired by vertically segmenting the input image and flipping the segments and the left and right regions of a third extended image.

Abstract: A substrate foreign matter inspection device includes: an image data obtaining device that obtains image data of a target inspection area in the printed circuit board including a printed portion of the solder paste; a storage that stores a neural network and a model, the model being generated by learning of the neural network that includes an encoding portion and a decoding portion by using, as learning data, only image data of the target inspection area that do not include any foreign matter; and a control device that obtains reconfigured image data by inputting original image data obtained by the image data obtaining device into the model, compares the original image data with the reconfigured image data, and determines whether any foreign matter is present or absent on the printed circuit board based on a result of comparison with the reconfigured image data.