Real-time document forgery detection

A method for detecting document forgery in real-time via a smart device, the method performed by a microprocessor running on the smart device, the method comprising:

• capturing, via a smart device, an image of a document;
• storing the image at the smart device for a pre-determined duration of time;
• extracting, via the smart device, data from the image;
• creating a dataset based on the extracted data, the dataset comprising:
• a document type for the document, the document being displayed, at least in part, on the image;
• and
• document details included in the document;
• retrieving, from a remote server, user account data associated with the user account, the user account data comprising profile data and account activity data, the user account data identified based on the document details;
• confirming, on the smart device, that the user account data retrieved from the remote server correlates to user account identifying data included in the document details;
• and
• following the confirming, determining a validity of the document by:
• feeding the dataset to a forged document detection engine running on the smart device;
• comparing, on the smart device, the dataset to machine learning (“ML”) data stored in a database at the smart device;
• when one or more discrepancies are identified between the dataset and the ML data, transmitting an electronic fraud alert notification from the smart device to a network of smart devices, the network of smart devices within a pre-determined proximity to the smart device;
• and
• when one or more discrepancies are not identified, validating the document, at the smart device;
• wherein:
• the determining of the validity of the document via the smart device improves a validity response time by executing the determination locally at the smart device and independent of communications between the smart device and the remote server.

A system for detecting document forgery in real-time leveraging edge computing, the system comprising:

• a first smart device associated with a first user, the first smart device from a network of smart devices;
• a second smart device associated with a second user, the second smart device from the network of smart devices, the second smart device being within a pre-determined proximity to the first smart device;
• a machine learning (“ML”) forged document detection engine running at the first smart device and at the second smart device;
• and
• a remote server, the remote server in electronic communication with each of the first smart device and the second smart device;
• wherein:
• when a document is positioned for scanning by the first smart device, the first smart device is configured to:
• capture a first image of the document;
• store the first image, for a pre-determined time period, at a data cache at the first smart device and at a data cache at the second smart device;
• extract, via the first smart device, data from the first image;
• create, via the first smart device, a dataset based on the extracted data, the dataset comprising:
• a document type for the document, the document being displayed, at least in part, on the first image;
• and
• document details included in the document;
• retrieve, from a remote server, user account data, the user account data comprising profile data and account activity data associated with the user account;
• confirm, on the first smart device, that the user account data retrieved from the remote server correlates to user account identifying data included in the document details;
• determine a first validity of the document by:
• feeding the dataset to the forged document detection engine running on the first smart device;
• comparing, on the first smart device, the dataset to ML data stored in a database at the smart device;
• identifying one or more discrepancies in between the dataset and the ML data;
• and
• in response to the identifying, transmitting an electronic fraud alert notification and the first image from the first smart device to the network of smart devices, the network of smart devices within a pre-determined proximity to the first smart device;
• and
• the second smart device is configured to, when a document is positioned for scanning by the second smart device:
• capture, via the second smart device, a second image of the document;
• prior to determining a second validity of the document, compare the second image to images stored in the data cache, the comparing including executing a determination to determine if the second image has a threshold level of similarity with an image stored in the data cache;
• determine, by the second smart device, that the second image has the threshold level of similarity with the first image;
• in response to the determining, suspend the second validity determination of the document;
• and
• transmit an instruction to the remote server to lock the user account;
• wherein:
• the determining of the validity of the document via the first smart device improves a validity response time by executing the determination locally at the first smart device and not requiring communications between the first smart device and the remote server.