Image-based authorization systems
Aspects described herein may allow transactions to be authenticated based on images taken at the transaction site. For example, a computing device may decline a transaction request based on a first category of the transaction indicated in the category information violates a transaction category restriction of the user account. The computing device may send, to a user device associated with the user account, an instruction to upload one or more photos that depict a physical environment where the transaction is requested. If the photos indicate the requested transaction belongs to another category that does not violate the transaction category restriction, the system may approve the transaction based on the photos.
1 . A method comprising:
receiving, by a first computing device and from a second computing device, a first message for authorization of a user request associated with a user account;
causing, on a graphical user interface on a user device associated with the user account, output of:
a plurality of user selectable options, each associated with a respective category; and
a prompt to take one or more photos that depict a physical environment where the user request is made;
receiving, from the user device:
the one or more photos; and
a selection, of one of the plurality of user selectable options, that indicates the user request belonging to a first category;
providing, as input to a trained machine learning model, at least a portion of the one or more photos;
receiving, as output from the trained machine learning model and in response to the input, an identification of one or more objects in the physical environment associated with the user request;
identifying, based on the one or more photos and the one or more objects, one or more times of a day when the one or more photos were taken;
determining, based on the one or more objects and the one or more times of the day, that the user request belongs to the first category; and
transmitting, to a third computing device and based on a determination that authorizing a user request of the first category does not violate a category restriction, a second message indicating an authorization of the user request.
2 . The method of claim 1 , wherein the second computing device comprises a smart card reader or a point of sale (POS) device.
3 . The method of claim 1 , wherein the prompt comprises a second request to take the one or more photos within a time range after the prompt is sent, and wherein the providing the one or more photos is based on determining that the one or more photos were received within the time range.
4 . The method of claim 1 , wherein the one or more photos each depict the physical environment from a different perspective, and wherein the identification of the one or more objects identifies a first object based on different perspectives of the first object.
5 . The method of claim 1 , further comprising:
receiving, from the user device, authentication information associated with the one or more photos; and
authenticating, based on the authentication information, that the one or more photos depict the physical environment where the user request is made.
6 . The method of claim 1 , further comprising:
training, using training data comprising a plurality of photos depicting a plurality of photos at different times of day, the machine learning model to output, in response to an input photo, a prediction of a time of day when the input photo was taken.
7 . A first computing device comprising:
one or more processors;
memory storing computer instructions that, when executed by the one or more processors, configure the first computing device to:
receive, from a second computing device, a first message for authorization of a user request associated with a user account;
cause, on a graphical user interface on a user device associated with the user account, output of:
a plurality of user selectable options, each associated with a respective category; and
a prompt to take one or more photos that depict a physical environment where the user request is made;
receive, from the user device:
the one or more photos; and
a selection, of one of the plurality of user selectable options, that indicates the user request belonging to a first category;
provide, as input to a trained machine learning model, at least a portion of the one or more photos;
receiving, as output from the trained machine learning model and in response to the input, an identification of one or more objects in the physical environment associated with the user request; and
identifying, based on the one or more photos and the one or more objects, one or more times of a day when the one or more photos were taken;
determine, based on the one or more objects and the one or more times of the day, that the user request belongs to the first category; and
transmit, to a third computing device and based on a determination that authorizing a user request of the first category does not violate a category restriction, a second message indicating an authorization of the user request.
8 . The first computing device of claim 7 , wherein the second computing device comprises a smart card reader or a point of sale (POS) device.
9 . The first computing device of claim 7 , wherein the prompt comprises a second request to take the one or more photos within a time range after the prompt is sent, and wherein the instructions, when executed by the one or more processors, configure the first computing device to provide the one or more photos based on determining that the one or more photos were received within the time range.
10 . The first computing device of claim 7 , wherein the one or more photos each depict the physical environment from a different perspective, and wherein the identification of the one or more objects identifies a first object based on different perspectives of the first object.
11 . The first computing device of claim 7 , wherein the instructions, when executed by the one or more processors, further configure the first computing device to:
receive, from the user device, authentication information associated with the one or more photos; and
authenticate, based on the authentication information, that the one or more photos depict the physical environment where the user request is made.
12 . The first computing device of claim 7 , wherein the instructions, when executed by the one or more processors, further configure the first computing device to train, using training data comprising a plurality of photos depicting a plurality of photos at different times of day, the machine learning model to output, in response to an input photo, a prediction of a time of day when the input photo was taken.
13 . A non-transitory computer-readable medium storing computer instructions that, when executed, cause performance of actions comprising:
receiving, by a first computing device from a second computing device, a first message for authorization of a user request associated with a user account;
causing, on a graphical user interface on a user device associated with the user account, output of:
a plurality of user selectable options, each associated with a respective category; and
a prompt to take one or more photos that depict a physical environment where the user request is made;
receiving, from the user device:
the one or more photos; and
a selection, of one of the plurality of user selectable options, that indicates the user request belonging to a first category;
providing, as input to a trained machine learning model, at least a portion of the one or more photos;
receiving, as output from the trained machine learning model and in response to the input, an identification of one or more objects in the physical environment associated with the user request;
identifying, based on the one or more photos and the one or more objects, one or more times of a day when the one or more photos were taken;
determining, based on the one or more objects and the one or more times of the day, that the user request belongs to the first category; and
transmitting, to a third computing device and based on a determination that authorizing a user request of the first category does not violate a transaction category restriction, a second message indicating an authorization of the user request.
14 . The non-transitory computer-readable medium of claim 13 , wherein the second computing device comprises a smart card reader or a point of sale (POS) device.
15 . The non-transitory computer-readable medium of claim 13 , wherein the prompt comprises a second request to take the one or more photos within a time range after the prompt is sent, and wherein the instructions, when executed, cause providing the one or more photos based on determining that the one or more photos were received within the time range.
16 . The non-transitory computer-readable medium of claim 13 , wherein the one or more photos each depict the physical environment from a different perspective, and wherein the identification of the one or more objects identifies a first object based on different perspectives of the first object.
17 . The non-transitory computer-readable medium of claim 13 , wherein the instructions, when executed, further cause performance of actions comprising:
receiving, from the user device, authentication information associated with the one or more photos; and
authenticating, based on the authentication information, that the one or more photos depict the physical environment where the user request is made.
18 . The non-transitory computer-readable medium of claim 13 , wherein the instructions, when executed, further cause training, using training data comprising a plurality of photos depicting a plurality of photos at different times of day, the machine learning model to output, in response to an input photo, a prediction of a time of day when the input photo was taken.