In ear device customization using machine learning

1. A method comprising:

storing a training set including anthropometric data of a plurality of different users, the anthropometric data for each of the plurality of different users comprising an image of ears of a given user from the plurality of users and a corresponding three-dimensional (3D) geometry of the ears of the given user that describes shapes of external portions of the ears and portions of ear canals of the given user;

training a machine learning algorithm using the training set including the anthropometric data of the plurality of users, the trained machine learning algorithm configured to generate 3D geometries of ears;

receiving anthropometric data of an ear of a user, the anthropometric data including at least one image describing an external portion of the ear of the user;

generating a 3D geometry of the ear of the user by applying the anthropometric data of the ear of the user that describes the external portion of the ear of the user to the trained machine learning algorithm, wherein the generated 3D geometry describes a shape of at least a portion of an ear canal of the ear of the user; and

generating a design of an in-ear device based in part on the generated 3D geometry of the ear of the user, the design describing a shell of the in-ear device that is customized to fit within at least a portion of the ear canal of the user.

2. The method of claim 1 , further comprising:

detecting, based on the anthropometric data of the user, features of the ear of the user; and

generating, based on the features of the ear of the user, the 3D geometry of the ear of the user.

3. The method of claim 2 , wherein the features of the ear of the user include at least one of an entrance of the ear canal of the user, a conchal bowl of the ear of the user, and a pinna of the ear of the user.

4. The method of claim 1 , wherein generating the 3D geometry of the ear of the user comprises:

determining features describing the ear of the user using the anthropometric data of the user;

identifying key points and detected edges describing the ear of the user using the machine learning algorithm and the features; and

forming the 3D geometry of the ear of the user based in part on the key points and the detected edges.

5. The method of claim 4 , further comprising:

predicting key points for a portion of the ear of the user that is occluded;

creating a shape basis for the portion of the ear of the user using the detected key points; and

wherein forming the 3D geometry of the ear of the user is based in part on the shape basis.

6. The method of claim 1 , wherein at least a portion of the shell of the in-ear device is configured to seal against the ear canal.

7. The method of claim 1 , wherein the shell of the in-ear device houses a speaker and two acoustic sensors.

8. The method of claim 1 , wherein the design of the in-ear device is provided to a manufacturer of the in-ear device.

9. The method of claim 1 , wherein the anthropometric data includes measurements of the ear of the user, and the measurements are used to generate the 3D geometry of the ear of the user.

10. The method of claim 1 , wherein the anthropometric data is an image of the ear of the user received from a device, and the image is used to generate the 3D geometry of the ear of the user.

11. The method of claim 1 , wherein the anthropometric data describing the ear of the user is determined from a video of the user.

12. A system comprising:

an imaging device configured to capture an image of an external portion of an ear of a user; and

a controller configured to:

store a training set including anthropometric data of a plurality of different users, the anthropometric data for each of the plurality of different users comprising an image of ears of a given user from the plurality of users and a corresponding three-dimensional (3D) geometry of the ears of the given user that describe shapes of external portions of the ears and portions of ear canals of the given user;

train a machine learning algorithm using the training set including the anthropometric data of the plurality of users, the trained machine learning algorithm configured to generate 3D geometries of ears;

receive anthropometric data of the ear of the user, the anthropometric data including the captured image of the external portion of the user that describes the external portion of the ear of the user;

generate a 3D geometry of the ear of the user by applying the anthropometric data within the image of the external portion of the ear to the trained machine learning algorithm, wherein the generated 3D geometry describes a shape of at least a portion of an ear canal of the ear of the user; and

generate a design of an in-ear device based in part on the generated 3D geometry of the ear of the user, the design describing a shell of the in-ear device that is customized to fit within at least a portion of the ear canal of the user.

13. The system of claim 12 , wherein the controller is further configured to:

detect, based on the anthropometric data of the user, features of the ear of the user; and

generate, based on the features of the ear of the user, the 3D geometry of the ear of the user.

14. The system of claim 13 , wherein the features of the ear of the user include at least one of an entrance of the ear canal of the user, a conchal bowl of the ear of the user, and a pinna of the ear of the user.

15. The system of claim 12 , wherein generating the 3D geometry of the ear of the user comprises:

determining features describing the anthropometric data of the user;

identifying key points and detected edges describing the ear of the user using the machine learning algorithm and the features;

compiling the key points and detected edges to form the 3D geometry of the ear of the user.

16. The system of claim 12 , wherein at least a portion of the shell of the in-ear device is configured to seal against the ear canal.

17. The system of claim 12 , wherein the shell of the in-ear device houses a speaker and an acoustic sensor.

18. The system of claim 12 , wherein the design of the in-ear device is provided to a manufacturer of the in-ear device.

19. The system of claim 12 , wherein the anthropometric data includes measurements of the ear of the user, and the measurements are used to generate the 3D geometry of the ear of the user.

20. A non-transitory computer readable storage medium comprising computer executable code that when executed by one or more processors causes the one or more processors to perform operations comprising:

storing a training set including anthropometric data of a plurality of different users, the anthropometric data for each of the plurality of different users comprising an image of ears of a given user from the plurality of users and a corresponding three-dimensional (3D) geometry of the ears of the given user that describe shapes of external portions of the ears and portions of ear canals of the given user;

training a machine learning algorithm using the training set including the anthropometric data of the plurality of users, the trained machine learning algorithm configured to generate 3D geometries of ears;

receiving anthropometric data of an ear of a user, the anthropometric data including at least one image describing an external portion of the ear of the user;

generating a 3D geometry of the ear of the user by applying the anthropometric data of the ear of the user that describes an external portion of the ear of the user to the trained machine learning algorithm, wherein the generated 3D geometry describes a shape of at least a portion of an ear canal of the ear of the user; and

generating a design of an in-ear device based in part on the generated 3D geometry of the ear of the user, the design describing a shell of the in-ear device that is customized to fit within at least a portion of the ear canal of the user.

21. The non-transitory computer readable storage medium of claim 20 , the one or more processors to perform operations further comprising:

detecting, based on the anthropometric data of the user, features of the ear of the user; and

generating, based on the features of the ear of the user, the 3D geometry of the ear of the user.