Abstract: Techniques for performing centralized unsuperivised learning in a multi-domain system are described. A user may request labeled data for an ML task, where the request includes a prompt for obtaining relevant explicit user feedback. The system may use the prompt to collect explicit user feedback for relevant runtime user inputs. After a duration of time (in the user's request for labeled data) has elapsed, the system determines whether collected user feedback indicates processing of the user input was defective and, if so, determines a cause of the defective processing. The system then uses one or more label generators to generate labeled data using the collected user feedback, whether the processing was defective, and the determined defect cause.

Abstract: A method for real-time dynamic 4D reconstruction can include detecting one or more topology changes between reconstructed frames and a new incoming frame by detecting a set of discontinuities in a first surface mesh associated with the reconstructed frames; duplicating cells of a first volumetric cell structure associated with the reconstructed frames at the set of discontinuities to generate a set of nodes, the set of nodes having a non-manifold connectivity; and fusing a depth image of the new incoming frame with the first volumetric cell structure having the set of nodes to form a next volumetric cell structure with the non-manifold connectivity. A next surface mesh extracted from the next volumetric cell structure can then be output for rendering a live frame.

Abstract: An electronic device for object rigging includes a processor. The processor is configured to obtain a 3D scan of an object. The processor is also configured to match a rigged parametric model to the 3D scan by minimizing a surface distance and 3D joint errors between the rigged parametric model and the 3D scan. The processor is further configured to identify a correspondence between the rigged parametric model and the 3D scan. Additionally, the processor is configured to transfer attributes of the rigged parametric model to the 3D scan based on the correspondence to generate a rigged 3D scan. The processor is also configured to apply animation motion to the rigged 3D scan. The rigged 3D scan with the applied animation motion is rendered on a display.

Abstract: A method for real-time dynamic 4D reconstruction can include detecting one or more topology changes between reconstructed frames and a new incoming frame by detecting a set of discontinuities in a first surface mesh associated with the reconstructed frames; duplicating cells of a first volumetric cell structure associated with the reconstructed frames at the set of discontinuities to generate a set of nodes, the set of nodes having a non-manifold connectivity; and fusing a depth image of the new incoming frame with the first volumetric cell structure having the set of nodes to form a next volumetric cell structure with the non-manifold connectivity. A next surface mesh extracted from the next volumetric cell structure can then be output for rendering a live frame.

Abstract: An electronic device for object rigging includes a processor. The processor is configured to obtain a 3D scan of an object. The processor is also configured to match a rigged parametric model to the 3D scan by minimizing a surface distance and 3D joint errors between the rigged parametric model and the 3D scan. The processor is further configured to identify a correspondence between the rigged parametric model and the 3D scan. Additionally, the processor is configured to transfer attributes of the rigged parametric model to the 3D scan based on the correspondence to generate a rigged 3D scan. The processor is also configured to apply animation motion to the rigged 3D scan. The rigged 3D scan with the applied animation motion is rendered on a display.

Abstract: An embodiment of the invention provides a 3D display device and a manufacturing method thereof. The 3D display device includes a display unit; and a birefringent grating stacked at a light emitting side of the display unit, wherein the birefringent grating including a concave lens film, and liquid crystal polymer covering the concave lens film, a long or short axis of the liquid crystal polymer being parallel to a ridge direction of the concave lens film, and a polarization direction of emitting light of the display unit is parallel to the ridge direction of the concave lens film of the birefringent grating.

Abstract: An embodiment of the invention provides a 3D display device and a manufacturing method thereof. The 3D display device comprises a display unit; and a birefringent grating stacked at a light emitting side of the display unit, wherein the birefringent grating comprising a concave lens film, and liquid crystal polymer covering the concave lens film, a long or short axis of the liquid crystal polymer being parallel to a ridge direction of the concave lens film, and a polarization direction of emitting light of the display unit is parallel to the ridge direction of the concave lens film of the birefringent grating.