Facial recognition for drivers
Methods for performing improving facial recognition of a driver in a vehicle are disclosed. A set of images is received. Each of the set of images includes a portion that is identified as a face. The identification is based on an application of a first machine-learned model to each of the set of images. The application of the first machine-learned model is performed by an application associated with a client camera device mounted in a vehicle. Based on a determination that the set of images matches one or more reference images stored in a database with a confidence level that is equal to or greater than a confidence threshold, a person corresponding to the one or more reference images is associated as a driver of the vehicle during a time period in which the set of images was captured.
1. A system comprising:
one or more memories;
one or more computer processors;
a set of instructions incorporated into the one or more memories, the set of instructions configuring the one or more processors to perform operations comprising:
receiving, at a server application executing on an application server, a set of images, each of the set of images including a portion that is identified as a face, the identification based on an application of a first machine-learned model to each of the set of images by a client application executing on a client device; and
based on a determination, at the server application, that an accuracy of the first machine-learned model is below an accuracy threshold, the determining of the accuracy based on an application of a second machine-learned model to each of the set of images by the server application, the second machine-learned model using more feature inputs than the first machine-learned model, retraining the first machine-learned model and communicating the retrained first machine-learned model to the client for upgrading of the first machine-learned model.
2. The system of claim 1 , wherein the client application is associated with a client camera device that is configured to capture the set of images when an event occurs, the event pertaining to operation of a vehicle.
3. The system of claim 2 , wherein the event is a starting of the vehicle.
4. The system of claim 2 , wherein the client application includes a configurable parameter for specifying how many images are to be captured by the client camera device after the event occurs.
5. The system of claim 4 , wherein the client application includes a configurable parameter for specifying a time period of separation between captures of the specified number of images.
6. The system of claim 5 , further comprising, based on the accuracy failing below the accuracy threshold, sending a communication to the client camera device to adjust the configurable parameter for specifying the time period of separation.
7. The system of claim 1 , wherein the client application reduces the size of the image by cropping the image to exclude a portion of the image that does not include an identified face.
8. The system of claim 1 , further comprising, based on a determination that the set of images does not match one or more reference images predicted by the second machine-learned model with a confidence level, retraining the second machine-learned model.
9. A method comprising:
receiving, at a server application executing on an application server, a set of images, each of the set of images including a portion that is identified as a face, the identification based on an application of a first machine-learned model to each of the set of images by a client application executing on a client device; and
based on a determination, at the server application, that an accuracy of the first machine-learned model is below an accuracy threshold, the determining of the accuracy based on an application of a second machine-learned model to each of the set of images by the server application, the second machine-learned model using more feature inputs than the first machine-learned model, retraining the first machine-learned model and communicating the retrained first machine-learned model to the client for upgrading of the first machine-learned model.
10. The method of claim 9 , wherein the client application is associated with a client camera device that is configured to capture the set of images when an event occurs, the event pertaining to operation of a vehicle.
11. The method of claim 10 , wherein the event is a starting of the vehicle.
12. The method of claim 10 , wherein the client application includes a configurable parameter for specifying how many images are to be captured by the client camera device after the event occurs.
13. The method of claim 12 , wherein the client application includes a configurable parameter for specifying a time period of separation between captures of the specified number of images.
14. The method of claim 13 , further comprising, based on the accuracy failing below the accuracy threshold, sending a communication to the client camera device to adjust the configurable parameter for specifying the time period of separation.
15. The method of claim 9 , wherein the application reduces the size of the image by cropping the image to exclude a portion of the image that does not include an identified face.
16. The method of claim 9 , further comprising, based on a determination that the set of images does not match one or more reference images predicted by the second machine-learned model with a confidence level, retraining the second machine-learned model.
17. A non-transitory computer-readable storage medium comprising a set of instructions that, when executed by one or more computer processors, cause the one or more computer processors to perform operations, the operations comprising:
receiving, at a server application executing on an application server, a set of images, each of the set of images including a portion that is identified as a face, the identification based on an application of a first machine-learned model to each of the set of images by a client application executing on a client device; and
based on a determination, at the server application, that an accuracy of the first machine-learned model is below an accuracy threshold, the determining of the accuracy based on an application of a second machine-learned model to each of the set of images by the server application, the second machine-learned model using more feature inputs than the first machine-learned model, retraining the first machine-learned model and communicating the retrained first machine-learned model to the client for upgrading of the first machine-learned model.
18. The non-transitory computer-readable storage medium of claim 17 , wherein the client application is associated with a client camera device that is configured to capture the set of images when an event occurs, the event pertaining to operation of a vehicle.
19. The non-transitory computer-readable storage medium of claim 18 , wherein the event is a starting of the vehicle.
20. The non-transitory computer-readable storage medium of claim 18 , wherein the client application includes a configurable parameter for specifying how many images are to be captured by the client camera device after the event occurs.