Gaze determination using glare as input
Machine learning systems and methods that learn glare, and thus determine gaze direction in a manner more resilient to the effects of glare on input images. The machine learning systems have an isolated representation of glare, e.g., information on the locations of glare points in an image, as an explicit input, in addition to the image itself. In this manner, the machine learning systems explicitly consider glare while making a determination of gaze direction, thus producing more accurate results for images containing glare.
1. A method of determining gaze direction in the presence of glare, the method comprising:
determining, using parallel processing circuitry, a gaze direction of a subject in an image, the gaze direction determined at least in part from output of one or more machine learning models having as inputs an isolated representation of glare in the image, one or more representations of at least a portion of a face of the subject, and a portion of the image corresponding to at least one eye of the subject, where the one or more representations of at least the portion of the face of the subject includes a first representation of the face of the subject including a mask of the face of the subject and a second representation of the face of the subject including a face grid of the face of the subject and the one or more machine learning models include a first machine learning model having as input the first representation of the face of the subject and a second machine learning model having as input the second representation of the face of the subject; and
initiating an operation based on the determined gaze direction.
2. The method of claim 1 , further comprising generating, using the processing circuitry and based at least in part on the image, the representation of glare, the one or more representations of at least the portion of the face of the subject, and the portion of the image corresponding to the at least one eye of the subject.
3. The method of claim 1 , wherein one of the machine learning models has as inputs both the isolated representation of glare and the representation of at least the portion of the face of the subject.
4. The method of claim 1 , wherein one of the machine learning models has as input the isolated representation of glare, and another one of the machine learning models has as input the representation of at least the portion of the face of the subject.
5. The method of claim 4 , wherein the one of the machine learning models comprises fully connected layers having as input the isolated representation of glare.
6. The method of claim 1 , wherein the image is generated using sensor data obtained using a sensor device corresponding to a vehicle, wherein the initiating further comprises initiating an operation of the vehicle based on the determined gaze direction.
7. The method of claim 1 :
wherein the portion of the image corresponding to the at least one eye of the subject further comprises a first portion of the image corresponding to a first eye of the subject, and a second portion of the image corresponding to a second eye of the subject; and
wherein the one or more machine learning models comprise a first machine learning model having as input the first portion of the image, and a second machine learning model having as input the second portion of the image.
8. The method of claim 1 , the isolated representation of glare in the image further comprises a binary mask of glare points extracted from the image.
9. A system for determining gaze direction in the presence of glare, the system comprising:
a memory; and
parallel processing circuitry configured to:
determine a gaze direction of a subject in an image, the gaze direction determined at least in part from output of one or more machine learning models having as inputs an isolated representation of glare in the image, one or more representations of at least a portion of a face of the subject, and a portion of the image corresponding to at least one eye of the subject, where the one or more representations of at least the portion of the face of the subject includes a mask of the face of the subject and a face grid of the face of the subject and the one or more machine learning models include a first machine learning model having as input the mask of the face of the subject and a second machine learning model having as input the face grid of the face of the subject; and
initiate an operation based on the gaze direction.
10. The system of claim 9 , wherein the parallel processing circuitry is further configured to generate the isolated representation of glare, the one or more representations of at least the portion of the face of the subject, and the portion of the image corresponding to the at least one eye of the subject, the isolated representation of glare, the one or more representations of at least the portion of the face of the subject, and the portion of the image corresponding to the at least one eye of the subject being generated based at least in part on the image.
11. The system of claim 9 , wherein one of the machine learning models has as inputs both the isolated representation of glare and the representation of at least the portion of the face of the subject.
12. The system of claim 9 , wherein one of the machine learning models has as input the isolated representation of glare, and another one of the machine learning models has as input the representation of at least the portion of the face of the subject.
13. The system of claim 12 , wherein the one of the machine learning models comprises fully connected layers having as input the isolated representation of glare.
14. The system of claim 9 , wherein the operation further comprises operation of a vehicle based at least in part on the gaze direction.
15. The system of claim 9 :
wherein the portion of the image corresponding to the at least one eye of the subject further comprises a first portion of the image corresponding to a first eye of the subject, and a second portion of the image corresponding to a second eye of the subject; and
wherein the one or more machine learning models comprise a first machine learning model having as input the first portion of the image, and a second machine learning model having as input the second portion of the image.
16. A method of determining gaze direction in the presence of glare, the method comprising:
receiving first input data indicative of locations of glare in an image;
receiving second input data representing at least a portion of the image corresponding to a subject represented in the image;
determining a gaze direction of the subject using parallel processing circuitry, the gaze direction being an output of one or more machine learning models using the first input data and the second input data, where the first input data includes a mask of the face of the subject and the second input data includes a face grid of the face of the subject; and
initiating an operation based on the gaze direction.
17. The method of claim 16 , wherein one of the machine learning models has as inputs the first input data and a third input data comprising at least one location of an eye of the subject in the image.
18. The method of claim 16 , wherein one of the machine learning models has as input the first input data, and another one of the machine learning models has as input at least one location of an eye of the subject in the image.
19. The method of claim 16 , wherein the operation further comprises operation of a vehicle based at least in part on the gaze direction.
20. The method of claim 16 , wherein the one or more machine learning models further have a third input data comprising a portion of the image corresponding to a first eye of the subject, and wherein the one or more machine learning models further have a fourth input data comprising a portion of the image corresponding to a second eye of the subject.
21. The method of claim 16 , wherein the one or more machine learning models further have a third input data comprising a representation of a face in the image.