An image capture device may capture two hemispherical views of a scene. The two hemispherical view of the scene may be stitched along a stitch line. The image capture device may be rotated to align the stitch line with a mid-line of a panoramic field of view of the scene. Separate exposure settings may be used to capture the two hemispherical views of the scene, with the exposure settings increasing the dynamic range of the scene depicted within the panoramic field of view of the scene. The panoramic field of view of the scene may be punched out as panoramic visual content.

A head-mounted wearable device includes a frame mountable on a head of a user; an infrared imaging device arranged to image a face of the user when the frame is mounted on the head of the user; and a computing system configured to perform operations including causing the infrared imaging device to capture an image of the face of the user using infrared light received at the infrared camera and initiating a biometric authentication process based on the image. The head-mounted wearable device may include a visible-light imaging device to image the face of the user with the computing system configured to perform operations including causing the visible-light imaging device to capture a second image of the face of the user using visible light received at the visible-light imaging device, with the biometric authentication process being based in part on the second image.

A method for efficient image registration between two images in the presence of inlier-poor domains includes receiving a set of candidate correspondences between the two images. An approximate homography between the two images is generated based upon a first correspondence in the correspondences. The set of candidate correspondences is filtered to identify inlier correspondences based upon the approximate homography. A candidate homography is computed based upon the inlier correspondences. The candidate homography can be selected as a final homography between the two images based upon a support of the candidate homography against the set of candidate correspondences. An image registration is performed between the two images based upon the candidate homography being selected as the final homography.

A method, performed by an electronic device, of generating a three-dimensional (3D) image, includes: obtaining a first image through a first camera of the electronic device and obtaining a second image through a second camera of the electronic device; obtaining depth information of a pixel included in the first image; identifying, based on the depth information, a first layer image and a second layer image from the first image; inpainting, based on the first image and the second image, at least a part of the first layer image; and generating, based on the second layer image and the inpainted first layer image, the 3D image including a plurality of layers.

In an approach for image registration performance assurance by optimizing system configurations, a processor evaluates alignment of a registered image and a fixed image using a pre-trained learning model. The registered image is generated with a first registration method. A processor provides a reward score to the alignment, the reward score being defined as a higher score indicating a better alignment. A processor generates a registration status represented as a feature vector that contains information about how the registered and fixed images are aligned. A processor determines a second registration method based on the reward score, the feature vector, and the first registration method.

Spherical rotation is described for encoding a video that has a wide field of view, such as a spherical or hemispherical video. One example relates to receiving encoded video including rotation orientation metadata, decoding the video, extracting the rotation orientation metadata, rotating the decoded video based on the rotation orientation metadata, generating a view of the rotated decoded video, and buffering the generated view for display.

Rectilinear viewport extraction from a region of a wide field of view is described using messaging in the video transmission field. Some embodiments pertain to a method that includes receiving a wide field of view video, receiving position data from a remote coupled user device, selecting a region of interest in the video based on the received position data, extracting the region of interest from the video, generating region metadata describing the extracted region of interest, encoding the extracted region of interest of the video, and transmitting the encoded video with the region metadata to the user device.

An example apparatus for adjusting eye gaze in images one or more processors to execute instructions to bidirectionally train a neural network; access a target angle and an input image, the input image including an eye in a first position; generate a vector field with the neural network; and generate a gaze-adjusted image based on the vector field, the gaze-adjusted image including the eye in a second position.

An example computing system is configured to extract gridline information from a two-dimensional drawing file and determine, for the gridline information, first coordinate information that is based on a first datum. The computing system converts the first coordinate information into second coordinate information that is based on a second datum, where the second coordinate information is used by a three-dimensional drawing file. The computing system is also configured to receive a request to generate a two-dimensional view of the three-dimensional drawing file, where the two-dimensional view includes an intersection of two meshes within the three-dimensional drawing file. The computing device generates the two-dimensional view of the three-dimensional drawing file and adds, to the generated two-dimensional view, (i) at least one gridline corresponding to the gridline information and (ii) dimensioning information involving the at least one gridline and at least one of the two meshes.

A computer-implemented method of processing video data comprising a first sequences of image frames containing a first instance of an object. The method includes isolating said first instance of the object within the first sequence of image frames, determining, using the isolated first instance of the object, first parameter values for a synthetic model of the object, modifying the first parameter values for the synthetic model of the object, rendering a modified first instance of the object using a trained machine learning model and the modified first parameter values for the synthetic model of the object, and replacing at least part of the first instance of the object within the first sequence of image frames with a corresponding at least part of the modified first instance of the object.