US 2019/0095698 A1 · Face Reconstruction from a Learned Embedding

Abstract

The present disclosure provides systems and methods that perform face reconstruction based on an image of a face. In particular, one example system of the present disclosure combines a machine-learned image recognition model with a face modeler that uses a morphable model of a human's facial appearance. The image recognition model can be a deep learning model that generates an embedding in response to receipt of an image (e.g., an uncontrolled image of a face). The example system can further include a small, lightweight, translation model structurally positioned between the image recognition model and the face modeler. The translation model can be a machine-learned model that is trained to receive the embedding generated by the image recognition model and, in response, output a plurality of facial modeling parameter values usable by the face modeler to generate a model of the face.

Claims21

Dependent claims

A computer-implemented method to obtain facial attribute data of a face, the method comprising:

obtaining, by one or more computing devices, an image of a face;
obtaining, by the one or more computing devices, an embedding supplied by a machine-learned image recognition model, the embedding supplied by the machine-learned image recognition model in response to receipt of the image of the face by the image recognition model;
inputting, by the one or more computing devices, the embedding into a machine-learned translation model;
and
receiving, by the one or more computing devices, a plurality of facial modeling parameter values as an output of the translation model, the plurality of facial modeling parameter values descriptive of a plurality of facial attributes of the face.

2 more elements

A computing system configured to obtain facial attribute data from an image of a face, the computing system comprising:

at least one processor;
a machine-learned translation model that is operable to receive an embedding obtained from a machine-learned image recognition model and, in response to receipt of the embedding, output a plurality of facial modeling parameter values that are descriptive of a plurality of facial attributes of the face;
at least one non-transitory computer-readable medium that stores instructions that, when executed by the at least one processor, cause the computing system to:
obtain the embedding from the image recognition model, the embedding produced by the image recognition model in response to receipt of the image of the face by the image recognition model;
input the embedding into the translation model;
and
receive the plurality of facial modeling parameter values as an output of the translation model.

4 more elements

Dependencies are inferred from the claim text.

Face Reconstruction from a Learned Embedding

Life of this patent

Face Reconstruction from a Learned Embedding

Life of this patent

Abstract

Claims21