The present disclosure relates to systems, methods, and non-transitory computer-readable media that generate preliminary object masks for objects in an image, surface the preliminary object masks as object mask previews, and on-demand converts preliminary object masks into refined object masks. Indeed, in one or more implementations, an object mask preview and on-demand generation system automatically detects objects in an image. For the detected objects, the object mask preview and on-demand generation system generates preliminary object masks for the detected objects of a first lower resolution. The object mask preview and on-demand generation system surfaces a given preliminary object mask in response to detecting a first input. The object mask preview and on-demand generation system also generates a refined object mask of a second higher resolution in response to detecting a second input.

A system, device and method of imaging using a real-time, AI-enabled analysis device coupled to an imaging device during an image scan of a subject includes: receiving data corresponding to a plurality of image frames from the imaging device and user input identifying a region of interest (ROI) in a first image frame; providing data corresponding to the first image frame, including the identified ROI and data corresponding to the remaining image frames to the AI-enabled data processing system; accepting a plurality of valid image frames from the plurality of image frames based on a predefined set of computer vision rules and a minimum accepted frame threshold; calculating, frame by frame, an ROI function value of the plurality of valid image frames; determining whether a predetermined ROI function value has been reached; and alerting an operator of the imaging device that the predetermined ROI function value has been reached.

System and method for generating acupuncture points on reconstructed 3D human body mesh for physical treatment are disclosed. The computing device obtains a first two-dimensional image of a human subject that captures at least a predefined portion of the human subject; processes the first two-dimensional image of the first human subject using a trained human body recovery model to obtain a plurality of parameters representing a three-dimensional human body mesh with corresponding acupuncture points. The trained human body recovery model includes an iterative three-dimensional regression module that is supervised by a discriminator and that minimizes a combined loss below a preset threshold. The combined loss includes a discriminator error that provides a measure of whether the obtained three-dimensional human body mesh with corresponding acupuncture points correspond to real human shape, pose, and acupuncture points.

A system for detecting and incorporating three-dimensional objects into a video stream reads an input video data stream. The user specifies areas of attention wherein said areas of attention or hotspots. Tracking movement of the hotspots generating a trajectory of said at least one object. Generating a cloud of points and tracking said points to detect configurations of points most similar to the initially defined hotspot. Obtaining a three dimensional topology defining a volume of interest in a three-dimensional space. Building virtual structures or pseudo objects that are placed within a spherical environment generated on the input video.

Embodiments include systems and methods for determining a 6D pose estimate associated with an image of a physical 3D object captured in a video stream. An initial 6D pose estimate is inferred and then further iteratively refined. The video stream may be frozen to allow the user to tap or touch a display to indicate a location of the user-input keypoints. The resulting 6D pose estimate is used to assist in replacing or superimposing the physical 3D object with digital or virtual content in an augmented reality (AR) frame.

Techniques are disclosed for aligning fiducial markers that commonly exist in each of multiple different N-dimensional (N-D) data sets. Notably, the N-D data sets are at least three-dimensional (3D) data sets. A first set and a second set of N-D data are accessed. A set of one or more fiducial markers that commonly exist in both those sets are identified. Based on the fiducial markers, one or more transformations are performed to align the two sets. Performing this alignment process results in at least a selected number of the common fiducial markers that exist in the two sets being within a threshold alignment relative to one another.

A data processing apparatus includes receiving circuitry to receive a plurality of images of an environment captured from respective different viewpoints, detection circuitry to detect a plurality of feature points in the plurality of captured images and to associate image information with each detected feature point indicative of an image property for a detected feature point, where each detected feature point represents a candidate landmark point for mapping the environment, selection circuitry to select one or more of the plurality of candidate landmark points, the one or more selected landmark points corresponding to a subset of the plurality of candidate landmark points, and mapping circuitry to generate, for the environment, a map including one or more of the selected landmark points, where each landmark point included in the map is defined by a three dimensional position and the associated image information for that landmark point.

Aspects of the present disclosure involve a system comprising a computer-readable storage medium storing at least one program, and a method for performing operations comprising: receiving a video that depicts a person. The operations further include identifying a set of skeletal joints of the person. The operations further include identifying a pose of the person depicted in the video based on positioning of the set of skeletal joints (or detecting a hand pose, detecting a mirror frame, or detecting a mobile device). The operations further include determining, based on the pose of the person (or detecting a hand pose, detecting a mirror frame, or detecting a mobile device), that the video comprises a mirror reflection of the person. The operations further include, in response to determining that the video comprises the mirror reflection of the person, causing display of a 3D virtual object in the video.

An authoring tool and method by which users (e.g., diagnosticians) are enabled to design, train, and deploy custom-made AI models tailored to their needs and specific to their data. In the approach herein, and using the authoring tool, users are provided the ability to provide (feed) actual labeling to the AI during the model training process itself (i.e., prior to validation testing of the model results themselves), preferably via a master template (or “questionnaire”) that is specific to a single modality-single body part pair.

A computer system maintains structure data indicating geometrical constraints for each structure category of a plurality of structure categories. The computer system generates a virtual 3D representation of a structure based on a set of images depicting the structure. For each image in the set of images, one or more landmarks are identified. Based on the landmarks, a candidate structure category is selected. The virtual 3D representation is generated based on the geometrical constraints of the candidate structure category and the landmarks identified in the set of images.