An Artificial Intelligence (AI) three-dimensional modeling system that analyzes and segments imagery of a room, generates a three-dimensional model of the room from the segmented imagery, identifies objects within the room, and conducts an assessment of the room based on the identified objects.

An augmented reality (AR) processing method, a computer readable storage medium, and an electronic device, relating to the technical field of AR. The AR processing method includes: obtaining a current frame image, extracting an image parameter of the current frame image, receiving information of a virtual object and displaying the virtual object; and editing the virtual object in response to an editing operation for the virtual object. The information of the virtual object corresponds to the image parameter of the current frame image and is determined by using a pre-stored mapping result.

A computer-implemented method for instance segmentation of at least one microscope image showing a plurality of objects, comprising: calculating positions of object centers of the objects in the microscope image; determining which image areas of the microscope image are covered by the objects; calculating Voronoi regions using the object centers as Voronoi sites; and determining an instance segmentation mask by separating the image areas covered by the objects into different instances using boundaries of the Voronoi regions.

Pixel image data of a scene is received in which the pixel image data includes a two-dimensional representation of an object in the scene. Point cloud data including three-dimensional point coordinates of a physical object within the scene corresponding to the two-dimensional representation of the object is received. The three-dimensional point coordinates include depth information of the physical object. The point cloud data is mapped to an image plane of the pixel image data to form integrated pixel image data wherein one or more pixels of the pixel image data have depth information integrated therewith. A three-dimensional bounding box is predicted for the object using a convolutional neural network based upon the integrated pixel image data.

An example method of capturing a 360° field-of-view image includes capturing, with one or more processors, a first portion of a 360° field-of-view using a first camera module and capturing, with the one or more processors, a second portion of the 360° field-of-view using a second camera module. The method further includes determining, with the one or more processors, a target overlap region based on a disparity in a scene captured by the first portion and the second portion and causing, with the one or more processors, the first camera module, the second camera module, or both the first camera module and the second camera module to reposition to a target camera setup based on the target overlap region. The method further includes capturing, with the one or more processors, the 360° field-of-view image with the first camera and the second camera arranged at the target camera setup.

A video generation method includes: acquiring an original image corresponding to a target frame, and identifying a target object in the original image; according to a sliding zooming strategy, performing sliding zooming processing on an initial background image in the original image excluding the target object, so as to obtain a target background image, wherein the sliding zooming strategy is at least used for indicating a sliding direction and a zooming direction of the initial background image, and the sliding direction is opposite to the zooming direction; according to the position of the target object in the original image, superimposing an image of the target object onto the target background image to obtain a target image corresponding to the target frame; and generating a target video on the basis of the target image corresponding to the target frame.

In some examples, an apparatus for mesh processing includes processing circuitry. The processing circuitry receives a first mesh frame with polygons representing a surface of an object, and determining that the first mesh frame is a non manifold type mesh in response to one or more singularity components in the first mesh frame. The processing circuitry converts the first mesh frame to a second mesh frame that is a manifold type mesh. The first mesh frame has first boundary loops that respectively correspond to second boundary loops in the second mesh frame. The processing circuitry detects the second boundary loops in the second mesh frame, and determines the first boundary loops in the first mesh frame according to the second boundary loops in the second mesh frame.

Vectorization of an image begins by receiving a two-dimensional rasterized image and returning a descriptor for each pixel in the image. Corner detection returns coordinates for all corners in the image. The descriptors are filtered using the corner positions to produce corner descriptors for the corner positions. A score matrix is extracted using the corner descriptors in order to produce a permutation matrix that indicates the connections between all of the corner positions. The corner coordinates and the permutation matrix are used to perform vector extraction to produce a machine-readable vector file that represents the two-dimensional image. Optionally, the corner descriptors may be refined before score extraction and the corner coordinates may be refined before vector extraction. A three-dimensional or N-dimensional image may also be input. A convolutional neural network performs descriptor extraction and corner detection; a graph neural network produces the refinements; and an optimal connection network performs score extraction.

A method for combining multiple images is disclosed herein. A target mapping matrix is determined based on a first image and a second image. The target mapping matrix is associated with a target correspondence between the first image and the second image. The first image and the second image are combined into a combined image based on the first target mapping matrix. The combined image is output by implementing the disclosed method.

Techniques are described for computing devices to perform automated operations to determine the acquisition locations of images, such as within a building interior based on automatically determined shapes of rooms of the building, and for using the determined image acquisition location information in further automated manners. The image may be a panorama image or of another type (e.g., a rectilinear perspective image) and acquired at an acquisition location in a multi-room building's interior, and the determined acquisition location for such an image may be at least a location on the building's floor plan and optionally an orientation/direction for at least a part of the image—in addition, the automated image acquisition location determination may be further performed without having or using information from any depth sensors or other distance-measuring devices about distances from an image's acquisition location to walls or other objects in the surrounding building.