System and method for detection of mobile device fault conditions

1. A method comprising:

installing an executable program on a mobile device;

prompting a user of the mobile device to place the mobile device so that a display section of the mobile device faces a reflective surface of a mirror;

performing a tracking calibration function to adjust the mobile device position and display for optimal image capture;

modifying a brightness setting and a camera sensitivity setting of the mobile device to optimize image capture; and

determining that the mobile device is in an acceptable orientation with respect to the mirror, and thereupon:

capturing an image of the mobile device as reflected from the mirror;

processing the captured image to determine whether a defect can be identified in the display of the mobile device, wherein the processing further comprises:

extracting, from the captured image, a subimage corresponding to a screen of the mobile device; and

transforming a perspective of the subimage into a rectangular aspect;

presenting transformed subimage to a pre-trained convolutional neural network (CNN) to identify one or more device defects; and

obtaining from the pre-trained neural network an indication of a first defect from the one or more device defects that corresponds to the screen of the mobile device, further including an indicia, created through class activation mapping functions, showing a portion of the mobile device screen that activates a defective class in the pre-trained neural network.

2. The method of claim 1 , further comprising formatting results of the determining step for transmission to a host server.

3. The method of claim 1 wherein the tracking calibration function further comprises:

retrieving one or more ambient lighting parameters from the mobile device;

adjusting a brightness parameter of the display of the mobile device based upon the retrieved ambient lighting parameters;

adjusting a sensitivity parameter of the camera of the mobile device from the retrieved ambient lighting parameters; and

determining from camera data obtained from the mobile device that the mobile device is out of optimal placement, whereupon:

a direction for the user to move the mobile device to obtain an optimal placement of the mobile device with respect to the mirror is calculated; and

the direction is displayed on the mobile device in a manner readable in the mirror by the user.

4. The method of claim 3 wherein adjusting a sensitivity parameter of the camera of the mobile device further comprises:

adjusting an ISO setting of the camera; and

adjusting a shutter speed of the camera.

5. The method of claim 3 wherein adjusting a sensitivity parameter of the camera of the mobile device further comprises adjusting an exposure compensation value associated with the camera of the mobile device.

6. The method of claim 3 further comprising:

displaying one or more fiducials on the display of the mobile device; and

sensing one or more of the fiducials from camera data obtained from the mobile device.

7. The method of claim 3 wherein: retrieving one or more ambient lighting parameters from the mobile device further comprises obtaining the parameters from an API associated with an operating system installed on the mobile device.

8. The method of claim 1 , wherein processing the captured image further comprises:

rotating the captured image to a portrait mode; and

cropping the rotated image.

9. The method of claim 8 , wherein:

the rotating and cropping steps are performed by a processor of the mobile device; and

the rotated and cropped image is further formatted for transmission to a host server.

10. The method of claim 1 , wherein processing the captured image to determine whether a defect can be identified in the display of the mobile device further comprises:

resampling the transformed subimage to a predetermined input image data configuration;

presenting the resampled transformed subimage to the pre-trained neural network to identify one or more device defects; and

obtaining from the pre-trained neural network an indication of a first defect from the one or more device defects that corresponds to the screen of the mobile device.

11. The method of claim 10 , wherein obtaining from the pre-trained neural network an indication of a first defect further comprises identifying a local area of a fault in the display of the mobile device using class activation mapping.

12. The method of claim 10 , further comprising:

receiving, by a server communicatively coupled to the mobile device, the captured image, and

wherein the processing of the captured image is performed within a processor of the server.

13. The method of claim 10 , further comprising training the pre-trained neural network with training data stored in a database in the server, the training data comprising at least: device screen training images and identified defects respectively corresponding to the device screen training images.

14. The method of claim 13 , wherein the training data stored in the database is augmented with a new training image corresponding to a captured image of the mobile device and a corresponding identified defect associated with the captured image of the mobile device.

15. The method of claim 14 , further comprising re-training the pre-trained neural network with the augmented training data.

16. The method of claim 10 , wherein the predetermined input image data configuration comprises a pixel configuration of 320×544 pixels.

17. The method of claim 10 , wherein the one or more device defects comprises one of:

a crack in the display of the mobile device;

one or more stuck pixels in the display of the mobile device;

one or more dead pixels in the display of the mobile device;

blemish or crack on a casing of the mobile device;

LCD bleed on the display of the mobile device;

a pink hue defect on the display of the mobile device; and

combinations thereof.