US 11,080,884 B2 · Point tracking using a trained network

Abstract

A trained network for point tracking includes an input layer configured to receive an encoding of an image. The image is of a locale or object on which the network has been trained. The network also includes a set of internal weights which encode information associated with the locale or object, and a tracked point therein or thereon. The network also includes an output layer configured to provide an output based on the image as received at the input layer and the set of internal weights. The output layer includes a point tracking node that tracks the tracked point in the image. The point tracking node can track the point by generating coordinates for the tracked point in an input image of the locale or object. Methods of specifying and training the network using a three-dimensional model of the locale or object are also disclosed.

Claims17

Dependent claims

A computer-implemented method for training a network for point tracking, comprising:

generating a three-dimensional model of at least a portion of a locale;
receiving a selection of a tracking point with reference to the three-dimensional model;
defining an output node in an output layer of the network as a first point tracking node associated with the tracking point;
applying to an input layer of the network:
(i) an encoding of a training image of the locale;
and (ii) a supervisor tracking point location on the training image, the supervisor tracking point location being a location of the tracking point on the training image;
generating, in response to the applying of the training image, a tracking point inference at the output node of the network using a set of internal weights of the network;
and
updating the set of internal weights based on a delta between the tracking point inference and the supervisor tracking point location.

6 more elements

Dependencies are inferred from the claim text.

Point tracking using a trained network

Life of this patent

Point tracking using a trained network

Life of this patent

Abstract

Claims17