Techniques for object identification

1 . A method comprising:

capturing, by a camera of a mobile device, an image, wherein the image includes a depiction of an authorization card;

generating, by the mobile device, a feature vector corresponding to the image, the feature vector comprising a respective numeric representation of a set of properties of the image;

providing, by the mobile device, the feature vector of the image as an input to a machine learning model, wherein the machine learning model is trained to;

assign a confidence score to the feature vector based on a position of the feature vector in an n-dimensional space relative to one or more known feature vectors, wherein each axis in the n-dimensional space corresponds to a respective property in the set of properties; and

output a predicted card label corresponding to the authorization card based on the feature vector;

receiving, by the mobile device, information that identifies the authorization card as an output from the machine learning model, the information that identifies the authorization card comprising the predicted card label;

comparing, by the mobile device, the information that identifies the authorization card to one or more criteria to verify the predicted card label outputted by the machine learning model; and

performing, by the mobile device, an operation based at least in part on the comparing of the information that identifies the authorization card to the one or more criteria.

2 . The method of claim 1 , wherein the operation comprises:

providing, by a speaker of the mobile device, a notification that identifies the authorization card.

3 . The method of claim 1 , wherein the one or more criteria include a target authorization card, and wherein comparing the information that identifies the authorization card to the one or more criteria comprises:

determining, by the mobile device, that the authorization card is not the target authorization card.

4 . The method of claim 3 , wherein the operation comprises:

providing, by a speaker of the mobile device, a notification comprising instructions to capture a second image that includes a depiction of a second authorization card.

5 . The method of claim 1 , wherein the one or more criteria includes one or more target locations for the authorization card, and wherein comparing the information that identifies the authorization card to the one or more criteria comprises:

determining, by the mobile device, a current location;

accessing, by the mobile device, the one or more target locations based at least in part on the information that identifies the authorization card;

comparing, by the mobile device, the current location and the one or more target locations; and

determining, by the mobile device, that the mobile device is not at a target location for the authorization card.

6 . The method of claim 5 , wherein the operation comprises:

providing, by a speaker of the mobile device, a notification comprising instructions to capture a second image that includes a depiction of a second authorization card.

7 . The method of claim 1 , wherein the training comprises:

collecting a set of authorization card images from a memory of the mobile device;

applying one or more transformations to each authorization card image including mirroring, rotating, smoothing, or contrast reduction to create a modified set of authorization card images;

creating a first training set comprising the collected set of authorization card images, the modified set of authorization card images, and a set of non-authorization card images;

training the machine learning model in a first stage using the first training set;

creating, by the mobile device, a second training set for a second stage of training comprising the first training set and non-authorization card images that are incorrectly detected as authorization card images after the first stage of training; and

training the machine learning model in a second stage using the second training set.

8 . A mobile device, comprising:

one or more memories; and

one or more processors in communication with the one or more memories and configured to execute instructions stored in the one or more memories to perform operations to:

capture, by a camera of the mobile device, an image, wherein the image includes a depiction of an authorization card;

generate, by the mobile device, a feature vector corresponding to the image, the feature vector comprising a respective numeric representation of a set of properties of the image;

provide, by the mobile device, the feature vector of the image as an input to a machine learning model, wherein the machine learning model is trained to;

assign a confidence score to the feature vector based on a position of the feature vector in an n-dimensional space relative to one or more known feature vectors, wherein each axis in the n-dimensional space corresponds to a respective property in the set of properties; and

output a predicted card label corresponding to the authorization card based on the feature vector;

receive, by the mobile device, information that identifies the authorization card as an output from the machine learning model, the information that identifies the authorization card comprising the predicted card label;

compare, by the mobile device, the information that identifies the authorization card to one or more criteria to verify the predicted card label outputted by the machine learning model; and

perform, by the mobile device, an operation based at least in part on the comparing of the information that identifies the authorization card to the one or more criteria.

9 . The mobile device of claim 8 , wherein the operation comprises operations to:

provide, by a speaker of the mobile device, a notification that identifies the authorization card.

10 . The mobile device of claim 8 , wherein the one or more criteria include a target authorization card, and wherein comparing the information that identifies the authorization card to the one or more criteria comprises operations to:

determine, by the mobile device, that the authorization card is not the target authorization card.

11 . The mobile device of claim 10 , wherein the operation comprises operations to:

provide, by a speaker of the mobile device, a notification comprising instructions to capture a second image that includes a depiction of a second authorization card.

12 . The mobile device of claim 8 , wherein the one or more criteria includes one or more target locations for the authorization card, and wherein comparing the information that identifies the authorization card to the one or more criteria comprises operations to:

determine, by the mobile device, a current location;

access, by the mobile device, the one or more target locations based at least in part on the information that identifies the authorization card;

compare, by the mobile device, the current location and the one or more target locations; and

determine, by the mobile device, that the mobile device is not at a target location for the authorization card.

13 . The mobile device of claim 12 , wherein the operation comprises operations to:

provide, by a speaker of the mobile device, a notification comprising instructions to capture a second image that includes a depiction of a second authorization card.

14 . The mobile device of claim 8 , wherein the training comprises operations to:

collect a set of authorization card images from a memory of the one or more memories of the mobile device;

apply one or more transformations to each authorization card image including mirroring, rotating, smoothing, or contrast reduction to create a modified set of authorization card images;

create a first training set comprising the collected set of authorization card images, the modified set of authorization card images, and a set of non-authorization card images;

train the machine learning model in a first stage using the first training set;

create a second training set for a second stage of training comprising the first training set and non-authorization card images that are incorrectly detected as authorization card images after the first stage of training; and

train the machine learning model in a second stage using the second training set.

15 . A non-transitory computer-readable medium storing a plurality of instructions that, when executed by one or more processors of a mobile device, cause the one or more processors to perform operations to:

capture, by a camera of the mobile device, an image, wherein the image includes a depiction of an authorization card;

generate, by the mobile device, a feature vector corresponding to the image, the feature vector comprising a respective numeric representation of a set of properties of the image;

provide, by the mobile device, the feature vector of the image as an input to a machine learning model, wherein the machine learning model is trained to:

assign a confidence score to the feature vector based on a position of the feature vector in an n-dimensional space relative to one or more known feature vectors, wherein each axis in the n-dimensional space corresponds to a respective property in the set of properties; and

output a predicted card label corresponding to the authorization card based on the feature vector;

receive, by the mobile device, information that identifies the authorization card as an output from the machine learning model, the information that identifies the authorization card comprising the predicted card label;

compare, by the mobile device, the information that identifies the authorization card to one or more criteria to verify the predicted card label outputted by the machine learning model; and

perform, by the mobile device, an operation based at least in part on the comparing of the information that identifies the authorization card to the one or more criteria.

16 . The non-transitory computer-readable medium of claim 15 , wherein the operation comprises operations to:

provide, by a speaker of the mobile device, a notification that identifies the authorization card.

17 . The non-transitory computer-readable medium of claim 15 , wherein the one or more criteria include a target authorization card, and wherein comparing the information that identifies the authorization card to the one or more criteria comprises operations to:

determine, by the mobile device, that the authorization card is not the target authorization card.

18 . The non-transitory computer-readable medium of claim 17 , wherein the operation comprises operations to:

provide, by a speaker of the mobile device, a notification comprising instructions to capture a second image that includes a depiction of a second authorization card.

19 . The non-transitory computer-readable medium of claim 15 , wherein the one or more criteria includes one or more target locations for the authorization card, and wherein comparing the information that identifies the authorization card to the one or more criteria comprises operations to:

determine, by the mobile device, a current location;

access, by the mobile device, the one or more target locations based at least in part on the information that identifies the authorization card;

compare, by the mobile device, the current location and the one or more target locations; and

determine, by the mobile device, that the mobile device is not at a target location for the authorization card.

20 . The non-transitory computer-readable medium of claim 19 , wherein the operation comprises operations to:

provide, by a speaker of the mobile device, a notification comprising instructions to capture a second image that includes a depiction of a second authorization card.