Photo relighting using deep neural networks and confidence learning

Patent metadata

Field	Value
Publication number	US-12136203-B2
Application number	US-202318236583-A
Country	US
Kind code	B2
Filing date	Aug 22, 2023
Priority date	Sep 24, 2018
Publication date	Nov 5, 2024
Grant date	Nov 5, 2024

Abstract

Apparatus and methods related to applying lighting models to images of objects are provided. A neural network can be trained to apply a lighting model to an input image. The training of the neural network can utilize confidence learning that is based on light predictions and prediction confidence values associated with lighting of the input image. A computing device can receive an input image of an object and data about a particular lighting model to be applied to the input image. The computing device can determine an output image of the object by using the trained neural network to apply the particular lighting model to the input image of the object.

First claim

What is claimed is:

1. A computer-implemented method, comprising:
receiving a training dataset comprising a plurality of images, wherein each image of the plurality of images is associated with a corresponding lighting model, wherein a given lighting model corresponding to a given image is indicative of a location of one or more environmental light sources with reference to an object in the given image;
training, based on the training dataset, a neural network by:
receiving, by a computing device, an input image and data about a target lighting model,
predicting an initial lighting model associated with the input image, and
predicting a relighting of the input image by replacing the initial lighting model with the target lighting model; and

providing the trained neural network.

2. The computer-implemented method of claim 1 , wherein the training is based on a cycle loss.

3. The computer-implemented method of claim 1 , wherein the training is based on an L2 loss measure.

4. The computer-implemented method of claim 1 , wherein the training is based on a log L1 loss measure.

5. The computer-implemented method of claim 1 , wherein the training utilizes a deep supervision technique to constrain one or more intermediate layers of the neural network.

6. The computer-implemented method of claim 1 , wherein the training is based on a generative adversarial net loss function.

7. The computer-implemented method of claim 1 , wherein the training utilizes confidence learning that is based on light predictions and prediction confidence values associated with lighting of the input image.

8. The computer-implemented method of claim 1 , wherein the object comprises a reflection property that diffusely reflects light.

9. The computer-implemented method of claim 1 , wherein the object comprises a face of a person.

10. The computer-implemented method of claim 1 , wherein the training of the neural network comprises training the neural network at the computing device.

11. The computer-implemented method of claim 1 , wherein the initial lighting model, the target lighting model, and the given lighting model, comprise data representing one or more of (i) a color, (ii) an intensity, (iii) an albedo, (iv) a light direction, (v) a surface normal, or (vi) one or more light sources, wherein at least one light source has a different location between the initial lighting model and the target lighting model.

12. The computer-implemented method of claim 1 , wherein the plurality of images comprise one or more objects under a plurality of different lighting conditions comprising one or more of (i) a first lighting provided from different directions, (ii) a second lighting provided of varying intensities, (iii) a third lighting provided with light sources of different colors, or (iv) a fourth lighting provided with different numbers of light sources.

13. A computer-implemented method, comprising:
receiving, by a computing device, an input image of an object and data about a target lighting model to be applied to the object;
predicting, by a trained neural network, (i) an initial lighting model indicative of a location of one or more environmental light sources with reference to the object in the input image, and (ii) a relighting of the object by applying the target lighting model, the neural network having been trained by:
receiving a given input image and input data about a given target lighting model,
predicting a given initial lighting model associated with the given input image, and
predicting a given relighting of the given input image by replacing the given initial lighting model with the given target lighting model; and

providing, by the computing device, an output image comprising the relighting of the object.

14. The computer-implemented method of claim 13 , wherein the object comprises a reflection property that diffusely reflects light.

15. The computer-implemented method of claim 13 , wherein the object comprises a face of a person.

16. The computer-implemented method of claim 13 , wherein the relighting of the object is modeled using the initial lighting model predicted by the trained neural network, and wherein the method further comprises:
providing the initial lighting model predicted by the trained neural network.

17. The computer-implemented method of claim 13 , wherein the providing of the output image comprises:
determining, by the computing device, a request to apply the target lighting model to the input image;
sending the request to apply the target lighting model to the input image from the computing device to a second computing device, the second computing device comprising the trained neural network; and
after sending the request, the computing device receiving, from the second computing device, the output image that applies the target lighting model to the input image.

18. The computer-implemented method of claim 13 , wherein the providing of the output image comprises:
obtaining the trained neural network at the computing device; and
determining the output image of the object by using the neural network as obtained.

19. The computer-implemented method of claim 13 , wherein the computing device comprises an image capturing device, and wherein the receiving of the input image comprises:
capturing the input image using the image capturing device.

20. A computing device, comprising:
one or more processors; and
data storage, wherein the data storage has stored thereon computer-executable instructions that, when executed by the one or more processors, cause the computing device to carry out functions comprising:
receiving, by the computing device, an input image of an object and data about a target lighting model to be applied to the object;
predicting, by a trained neural network, (i) an initial lighting model indicative of a location of one or more environmental light sources with reference to the object in the input image, and (ii) a relighting of the object by applying the target lighting model, the neural network having been trained by:
receiving a given input image and input data about a given target lighting model,
predicting a given initial lighting model associated with the given input image, and
predicting a given relighting of the given input image by replacing the given initial lighting model with the given target lighting model; and

providing, by the computing device, an output image comprising the relighting of the object.

Assignees

• Google Llc

Inventors

• Sun Tiancheng
• Tsai Yun-Ta
• Barron Jonathan

Classifications

• G06N3/08

  Learning methods · CPC title

• G06N3/048

  Activation functions · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06T15/50

  Lighting effects · CPC title

• G06T5/60

  using machine learning, e.g. neural networks · CPC title