Back to results
US 2024/0177359 A1Published application
Training a machine learning model for reconstructing occluded regions of a face
AssigneeSony Interactive Entertainment Inc.
InventorsRathish Krishnan, Deepali Arya, Manoj Srivastava, Seema Kataria
Life of this patent
Estimating the term…
Abstract
A machine learning (ML) model is trained using pairs of images. Each pair includes an image of a human face and a duplicate of the image with a computer game headset overlaid on the face using computer graphics. The ML model subsequently can be used to receive an image of a gamer wearing a headset and output a full-face image of the gamer for use in, e.g., social network settings related to the game.
Claims20
An apparatus comprising:
- at least one processor configured to:
- receive plural images of human faces with no occluded regions;
- for each of at least some of the images, generate a duplicate of the image with at least part of the face in the image occluded by a computer simulation headset to render a respective pair of images;
- and
- input the pairs of images to a machine learning (ML) model to train the model.
2 more elements
A device comprising:
- at least one computer storage that is not a transitory signal and that comprises instructions executable by at least one processor to:
- input plural pairs of images of people to at least one machine learning (ML) model to train the ML model;
- establish the pairs at least in part by receiving, for each pair, an original image of a human face and creating a second image of the original image partially occluded by a computer simulation headset.
A method comprising:
- receiving an image of a human face;
- generating an image of the human face with a computer-generated image of a computer simulation headset partially occluding the human face;
- and
- training at least one machine learning (ML) model using the image of the human face and the image of the human face with a computer-generated image of a computer simulation headset partially occluding the human face.
1 more element
Dependencies are inferred from the claim text.