Machine-learning models for image processing
Presented herein are systems and methods for the employment of machine learning models for image processing as may be performed by computing devices associated with an end user. A method may include obtaining video data comprising a plurality of frames including a document of a document type. The method may include executing an object recognition engine of a machine-learning architecture using image data of the plurality of frames, the object recognition engine trained to detect edges of documents. The method may include identifying, based on the edge detection, a plurality of boundaries for the document. The method may include validating, based on the plurality of boundaries, the document as the document type. The method may include transmitting via one or more networks, to a computer remote from the computing device, responsive to the validation of the type of document, the image data for the plurality of frames depicting the document.
1 . A method for client-side validation of document-imagery for remote processing, the method comprising:
obtaining, by a camera of a mobile client device associated with an end-user, video data comprising a plurality of frames having image data including a document and an operation request associated with the image data received via a user interface of the mobile client device;
extracting, by the mobile client device executing an object recognition engine, a set of environmental features from environmental imagery of the image data;
determining, by the mobile client device, a weighting factor for the set of environmental features based on a confidence level associated with the environmental features;
extracting, by the mobile client device executing the object recognition engine, a set of behavioral features associated with the end-user from the environmental imagery;
generating, by the mobile client device, a user validation score based upon the set of environmental features and the set of behavioral features, using a validating machine-learning model trained to identify a registered user according to training environmental features and training behavioral features and corresponding training labels, the user validation score indicating a likelihood that the end-user is the registered user; and
in response to validating the end-user as the registered user based on determining that the user validation score satisfies a user validation threshold, generating, by the mobile client device, an operation instruction for a backend server, the operation instruction including the image data of a frame and device metadata identifying the mobile client device.
2 . The method according to claim 1 , wherein the environmental features includes at least one of lighting characteristics, background consistency, or device orientation.
3 . The method according to claim 1 , wherein the behavioral features comprising at least one of device motion patterns, device handling posture, or temporal stability of the mobile client device during image capture.
4 . The method according to claim 1 , further comprising determining, by the mobile client device, a weighting factor for the set of behavioral features based on a confidence level associated with the environmental imagery.
5 . The method according to claim 1 , wherein the mobile client device updates one or more machine-learning models of the mobile client device at a periodic interval according to updated environmental features associated with a successful validation.
6 . The method according to claim 1 , wherein the mobile client device updates one or more machine-learning models of the mobile client device at a periodic interval according to updated behavioral features associated with a successful validation.
7 . The method according to claim 1 , further comprising, in response to determining that the user validation score does not satisfy the user validation threshold, executing, by the mobile client device, a remedial operation including at least one of: prompting the end-user for additional authentication data, halting the operation request, or disabling access to a remote deposit capture application.
8 . The method according to claim 1 , wherein the environmental imagery is obtained concurrently with document imagery during a check-imaging session.
9 . The method according to claim 1 , wherein the device metadata includes at least one of a device identifier, geolocation data, timestamp, or application version.
10 . A system for client-side validation of document-imagery for remote processing, the system comprising:
a mobile client device associated with an end-user, the mobile client device comprising at least one processor, a camera, wherein the mobile client device is configured to:
obtain, via the camera, video data comprising a plurality of frames having image data including a document and an operation request associated with the image data received via a user interface;
extract a set of environmental features from environmental imagery of the image data;
extract a set of behavioral features associated with the end-user from the environmental imagery;
determine a weighting factor for the set of behavioral features based on a confidence level associated with the environmental imagery;
generate a user validation score based upon the set of environmental features and the set of behavioral features, using a validating machine-learning model trained to identify a registered user according to training environmental features and training behavioral features and corresponding training labels, the user validation score indicating a likelihood that the end-user is the registered user; and
in response to validating the end-user as the registered user based on determining that the user validation score satisfies a user validation threshold, generate an operation instruction for a backend server, the operation instruction including the image data of a frame and device metadata identifying the mobile client device.
11 . The system of claim 10 , wherein the environmental features include at least one of lighting characteristics, background consistency, or device orientation.
12 . The system of claim 10 , wherein the behavioral features include at least one of device motion patterns, device handling posture, or temporal stability of the mobile client device during image capture.
13 . The system of claim 10 , wherein the mobile client device is further configured to determine a weighting factor for the set of environmental features based on a confidence level associated with the environmental imagery.
14 . The system of claim 10 , wherein the mobile client device is further configured to update one or more machine-learning models at a periodic interval according to updated environmental features associated with a successful validation.
15 . The system of claim 10 , wherein the mobile client device is further configured to update one or more machine-learning models at a periodic interval according to updated behavioral features associated with a successful validation.
16 . The system of claim 10 , wherein the mobile client device is further configured to, in response to determining that the user validation score does not satisfy the user validation threshold, execute a remedial operation including at least one of: prompting the end-user for additional authentication data, halting the operation request, or disabling access to a remote deposit capture application.
17 . The system of claim 10 , wherein the environmental imagery is obtained concurrently with document imagery during a check-imaging session.
18 . The system of claim 10 , wherein the device metadata includes at least one of a device identifier, geolocation data, timestamp, or application version.