Methods and systems for updating perception models based on geolocation features

1. A server comprising:

a controller programmed to:

obtain information about a first perception model installed in a vehicle moving from a first region to a second region;

determine a value of updating the first perception model;

determine whether the first perception model needs to be updated to a second perception model based on the value of updating the first perception model; and

transmit the second perception model to the vehicle in response to determining that the first perception model needs to be updated to the second perception model,

wherein the first perception model is pre-trained with geolocation-based features related to the first region,

wherein the second perception model is pre-trained with geolocation-based features related to the second region, and

wherein the value of updating the first perception model is determined based on difference between accuracy of identifying objects in the second region using the second perception model and accuracy of identifying the objects in the second region using the first perception model.

2. The server of claim 1 , wherein the value of updating the first perception model is determined further based on difference between processing time of an image using the second perception model and processing time of the image using the first perception model.

3. The server of claim 2 , wherein the value of updating the first perception model is determined further based on a cost of updating to the first perception model.

4. The server of claim 1 , wherein the first perception model detects a set of objects different from a set of objects detected by the second perception model.

5. The server of claim 1 , wherein the second perception model is pre-trained with geolocation-based features related to a new region.

6. The server of claim 1 , wherein the controller is further programmed to:

obtain a user profile of the vehicle; and

optimize the second perception model based on the user profile.

7. The server of claim 1 , wherein the controller is further programmed to update the second perception model based on at least one of time information, weather information, traffic incident information, and natural disaster information.

8. A method performed by a controller comprising:

obtaining information about a first perception model installed in a vehicle moving from a first region to a second region;

determining a value of updating the first perception model;

determining whether the first perception model needs to be updated to a second perception model based on the value of updating the first perception model; and

transmitting the second perception model to the vehicle in response to determining that the first perception model needs to be updated to the second perception model,

wherein the first perception model is pre-trained with geolocation-based features related to the first region,

wherein the second perception model is pre-trained with geolocation-based features related to the second region, and

wherein the value of updating the first perception model is determined based on difference between accuracy of identifying objects in the second region using the second perception model and accuracy of identifying the objects in the second region using the first perception model.

9. The method of claim 8 , wherein the value of updating the first perception model is determined further based on difference between processing time of an image using the second perception model and processing time of the image using the first perception model.

10. The method of claim 9 , wherein the value of updating the first perception model is determined further based on a cost of updating to the first perception model.

11. The method of claim 8 , wherein the first perception model detects a set of objects different from a set of objects detected by the second perception model.

12. The method of claim 8 , wherein the second perception model is pre-trained with geolocation-based features related to a new region.

13. A vehicle comprising:

a controller programmed to:

collect information about a first perception model of the vehicle moving from a first region to a second region;

transmit the information about the first perception model to a server;

receive a second perception model generated based on a value of updating the first perception model; and

update the first perception model to the second perception model,

wherein the controller is programmed to offload a perception task to the server while updating the first perception model to the second perception model,

wherein the first perception model is pre-trained with geolocation-based features related to the first region,

wherein the second perception model is pre-trained with geolocation-based features related to the second region, and

wherein the value of updating the first perception model is determined based on difference between accuracy of identifying objects in the second region using the second perception model and accuracy of identifying the objects in the second region using the first perception model.

14. The vehicle of claim 10 , wherein the value of updating the first perception model is determined further based on difference between processing time of an image using the second perception model and processing time of the image using the first perception model.

15. The vehicle of claim 14 , wherein the value of updating the first perception model is determined further based on a cost of updating to the first perception model.

16. The vehicle of claim 13 , wherein the first perception model detects a set of objects different from a set of objects detected by the second perception model.