Abstract: A computer system may identify a first position of a physical camera corresponding to a first time period. The computer system may render a first virtual scene for the first time period. The system may project the first scene onto a display surface to determine a first rendered image for the first time period. The computer system may receive a first camera image of the display surface from the camera during the first time period. The system may determine a first corrected position of the camera by comparing the first rendered image to the first camera image. The system may predict a second position of the camera corresponding to a second time period. The computer system may render a second virtual scene for the second time period. The system may project the second virtual scene onto the display surface to determine a second rendered image for the second time period.

Abstract: Embodiments of the present disclosure relate to training-free consistent text-to-image generation. A pre-trained text-to-image diffusion model is leveraged to generate images depicting a consistent subject for diverse prompts describing scenes. Inputs to the model are a text description of at least one subject with prompts (scene text descriptions) describing scenes, where each prompt is associated with a different generated image and the text description is used for all images that depict the subject. Internal activations (intermediate data) computed by the model during generation of the different images are shared for generation of the different images. A subject-driven shared attention block and correspondence-based feature injection are incorporated into the model to promote subject consistency within each image and/or between images. Additionally, layout diversity is encouraged while maintaining subject consistency.

Abstract: The present disclosure relates to a method and apparatus for representing dynamic neural radiance fields from unsynchronized videos. A method of acquiring a video at an arbitrary viewpoint based on a dynamic neural radiance fields model according to an embodiment of the present disclosure may comprise: inputting one or more videos acquired from one or more views for one scene into the dynamic neural radiance fields model; inputting a time embedding for the one or more videos into the dynamic neural radiance fields model; and rendering the video at the arbitrary viewpoint based on color information and density information output by the dynamic neural radiance fields model. Herein, time synchronization related to the time embedding may be performed by applying an individual time offset learned for each view, for the one or more views.

Abstract: A method for decoding three-dimensional (3D) content comprising extracting a video frame from a video, wherein the video frame includes connectivity information associated with the 3D content; and reconstructing the 3D content based on the connectivity information. The reconstructing comprises: determining an end of the connectivity information with respect to a mesh frameblock based on a termination connectivity coding unit (CCU) in the block; and determining an end of the connectivity information with respect to a mesh frame based on a termination block in the mesh frame.

Abstract: Techniques for performing an automated clash detection analysis on two-dimensional (2D) drawings associated with a given location of a construction project involve obtaining a set of 2D drawings for a construction project and identifying a subset of the 2D drawings that are associated with the given location of the construction project. From the subset of 2D drawings, a first 2D drawing and a second 2D drawing are selected for inclusion in the automated clash detection analysis. Based on respective sets of key points, the first and second 2D drawings are aligned and layered to produce an overlaid view. Objects in each 2D drawing are identified, and the overlaid view is analyzed to identify clashes between objects in the first 2D drawing and objects in the second 2D drawing. Respective visual representations of identified clashes may be displayed for user interaction.

Abstract: Methods, storage media, and systems for generating a three-dimensional line segment are disclosed. Exemplary implementations may: receive a plurality of images; generate a point cloud based on the plurality of images; detect a two-dimensional line segment in a first image; project a set of 3d points of the plurality of 3d points as 2d points in the first image; select projected 3d points that are proximate to 2d points along the 2d line segment; and generate a 3d line segment by connecting 3d points of the point cloud represented by the selected projected 3d points.

Abstract: Systems, methods, and apparatuses described herein may provide image processing, including displaying, by a mobile device, an image of an object located perpendicular to a reference object, calculating, based on at least one depth measurement determined using a depth sensor in the mobile device, the predicted height of the mobile device when the image was captured, calculating scale data for the image based on the predicted height, determining a reference line identifying the location of the object and the reference object in the image, segmenting pixels in the object in the image from pixels in the image outside the object, measuring the object based on the pixels in the object and the scale data, and generating model data comprising the object, the scale data, and the measurements.

Abstract: The present invention relates to a method for characterizing a zone of land, referred to as the target zone, intended for the installation of photovoltaic panels, the method comprising: receiving a bird's eye image of a target zone, detecting, in the image, a shadow cast over the target zone originating from an obstacle, determining a dimension, referred to as the main dimension, of the detected cast shadow, determining the position of the sun during the acquisition of the image on the basis of characteristics relating to the image, determining the height of the obstacle on the basis of the main dimension and the determined position of the sun, and characterizing the target zone on the basis of the height determined for the obstacle.

Abstract: A computer implemented method includes obtaining a plurality of 2D images of an object implanted in a patient and obtaining a computer-aided design (CAD) model corresponding to the object. The method further includes augmenting a 3D scan of the patient to provide a representation of the CAD model with a pose in a reference system of the 3D scan determined based on pose of the object in the 2D images.

Abstract: In various examples, a visualization of an environment may be generated using a Panini projection that is optimized based on detected scene content. For example, image data of an environment may be perspective projected (e.g., using a rectilinear projection) to generate a reference projection image, which may be analyzed to detect the presence of vanishing points and/or horizontal lines (e.g., in a central region). The image data of the environment may be projected using a Panini projection that is optimized based on distances to detected objects, the absence of a detected vanishing point, and/or the presence of a detected horizontal line to generate a Panini projection image. In some embodiments, vertical compression is applied to the Panini projection image to correct for distortion of horizontal lines (e.g., based on the presence of a detected horizontal line).

Abstract: A method includes obtaining an image that includes a plurality of pixels. The method includes identifying a first subset of the plurality of pixels that defines a particular geometry that a user of the electronic device is currently focusing on. The method includes warping the first subset of the plurality of pixels according to a first warping function that is based on a movement of the electronic device. The method includes warping a second subset of the plurality of pixels, that is different from the first subset, according to a second warping function that is different from the first warping function.

Abstract: An example of an information processing system generates a display mesh based on voxel data. The information processing system generates a voxel update range in a virtual space, based on game processing, and updates at least one of a density and a material ID, in the voxel data, for each of voxels corresponding to the voxel update range. The information processing system updates the display mesh corresponding to the updated voxel data. The information processing system, based on collision determination between a determination mesh and a determination shape corresponding to a determination target, generates an in-game behavior according to a material of the determination mesh at a collision position. The information processing system renders a virtual space including the display mesh, based on a texture corresponding to the material of the display mesh.

Abstract: An apparatus includes a communication interface configured to receive a compressed bitstream including a base mesh sub-bitstream and a processor operably coupled to the communication interface. The processor is configured to decode a plurality of submeshes from the base mesh sub-bitstream. The processor is also configured to subdivide a submesh of the plurality of submeshes according to a subdivision iteration count to generate at least one subdivided submesh, including determine a number of vertex positions for the at least one subdivided submesh by use of a number of vertices associated with an original submesh and by use of distortion information between the original submesh and a base mesh at each subdivision iteration associated with the subdivision iteration count. The processor is also configured to reconstruct at least a portion of a mesh-frame using the vertex positions corresponding to the at least one subdivided submesh.

Abstract: A system may perform motion capture using motion capture targets with concave reflector structures. For example, the motion capture target include a target body and a plurality of tracking markers located on respective portions of the surface of the target body. At least one tracking marker of the plurality of tracking markers may be a concave reflector structure including a tapered hole in the surface of the target body and at least a portion of a surface of the tapered hole may be reflective.

Abstract: Input images are obtained to include color data and depth data of a target. An image graph is generated based on the color data of the input images such that the image graph includes correlation values between the input images. A registration order is determined based on the image graph. Pose information of each input image is sequentially generated based on the registration order and the depth data of the input images. The input images are sequentially registered as registered images such that each registered image includes the pose information. A three-dimensional model of the target is reconstructed based on the registered images. The pose information may be generated and the elaborate three-dimensional model may be reconstructed by determining the registration order based on the color data and sequentially generating the pose information of the input images according to the registration order using the registered images.

Abstract: A system for video-based tracking of objects in one or more regions of interest. The system includes one or more cameras to capture image streams of the one or more regions of interest, the imaging streams including zero or more objects. A plurality of image processors of the system receive the captured image streams from the one or more cameras, and detect the one or more objects or object parts in the captured image streams and generate geometric and tracking data for detected objects. A fusion processor of the system receives the captured image streams and geometric and tracking data for detected objects from the plurality of image processors, and generates fused 3D referenced data using the detection and tracking data detected objects.

Abstract: A method includes obtaining a 2D image captured using an imaging sensor. The 2D image is associated with an imaging sensor pose. The method also includes providing the 2D image, the imaging sensor pose, and one or more additional imaging sensor poses to at least one machine learning model that is trained to generate a texture map and a depth map for the imaging sensor pose and each additional imaging sensor pose. The method further includes generating a stereo image pair based on the texture maps and the depth maps. The stereo image pair represents a 2.5D view of the 2D image. The 2.5D view includes a pair of images each including multiple collections of pixels and, for each collection of pixels, a common depth associated with the pixels in the collection of pixels. In addition, the method includes initiating display of the stereo image pair on an XR device.

Abstract: A deep learning network can perform three-dimensional (3D) image reconstruction of a scene from multi-view calibrated two-dimensional (2D) images. The network can include a convolutional neural network that performs feature extraction to generate feature pyramids corresponding to features at different levels of resolution. The feature pyramids can be used to compute a respective cost volume for each feature pyramid at each level of resolution, with the cost volume incorporating a learnable parameter that corresponds to a weight allocated to the “reference” feature pyramid relative to other feature pyramids. A depth map for each input image can be generated based at least in part on the cost volume.

Abstract: A method and a device of designing a metasurface, and a storage medium are provided. The method includes determining a phase distribution of an initial metasurface according to a preset image and a light source function of a first light source; the preset image includes a plurality of projection points in an array; performing a simulation based on the phase distribution of the initial metasurface and a second light source, so as to obtain a speckle image; if the speckle image does not meet a preset standard, optimizing the phase distribution of the initial metasurface through adjusting the projection points on the preset image, so as to determine a target phase distribution; generating the metasurface according to the target phase distribution.

Abstract: Systems and techniques are provided for generating a three-dimensional (3D) facial model. For example, a process can include obtaining at least one input image associated with a face. In some aspects, the process can include obtaining a pose for a 3D facial model associated with the face. In some examples, the process can include generating, by a machine learning model, the 3D facial model associated with the face. In some cases, one or more parameters associated with a shape component of the 3D facial model are conditioned on the pose. In some implementations, the 3D facial model is configured to vary in shape based on the pose for the 3D facial model associated with the face.