Learning-based 3D property extraction
AssigneeMangolytics, Inc.
InventorsSwupnil Kumar Sahai, Richard Hsu, Adith Balamurugan, Neel Sesh Ramachandran
Life of this patent
Estimating the term…
Abstract
Learning-based 3D property extraction can include: capturing a series of live 2D images of a participatory event including at least a portion of at least one reference visual feature of the participatory event and at least a portion of at least one object involved in the participatory event; and training a neural network to recognize at least one 3D property pertaining to the object in response to the live 2D images based on a set of timestamped 2D training images and 3D measurements of the object obtained during at least one prior training event for the neural network.
Claims20
A mobile device, comprising:
- a camera that captures a 2D image of a participatory event including at least a portion of a reference visual feature of the participatory event and at least a portion of an object of interest in the participatory event;
- and
- a neural network trained to extract from the 2D image a 3D property of the object within a 3D space of the participatory event by correlating an arrangement of pixels in the 2D image that correspond to the object and an arrangement of pixels in the 2D image that correspond to the reference visual feature to the 3D property.
A system for 3D property extraction, comprising:
- a single camera positioned to capture a live video of a participatory event including at least a portion of an object of interest in the participatory event and at least a portion of a reference visual feature of the participatory event;
- and
- a computing mechanism that implements a neural network trained to recognize a 3D property of the object within a 3D space of the participatory event in response to at least one 2D image in the live video by correlating an arrangement of pixels in the 2D image that correspond to the object and an arrangement of pixels in the 2D image that correspond to the reference visual feature to the 3D property.
A method for 3D property extraction, comprising:
- deploying a single camera positioned to capture a live video of a participatory event including at least a portion of an object of interest in the participatory event and at least a portion of a reference visual feature of the participatory event;
- and
- providing a neural network trained to recognize a 3D property of the object within a 3D space of the participatory event in response to at least one 2D image in the live video by correlating an arrangement of pixels in the 2D image that correspond to the object and an arrangement of the pixels in the 2D image that correspond to the reference visual feature to the 3D property.
Dependencies are inferred from the claim text.