Remote Check Deposit Limit Remediation With Active OCR
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital document verification systems face challenges in determining whether a document meets institutional requirements quickly and in real-time, and users are unable to modify requests or schedules due to remote deposit limits, leading to frustration and potential fraud.
Innovation Solution
Implementing active OCR on a client device to extract check data in real-time, combined with machine learning models, to determine available deposit splits and schedules that adhere to institutional limits, allowing users to select and schedule deposits without extensive image processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If detailed image processing is performed to determine whether a document meets institutional requirements, then measurement precision is improved, but loss of time increases and productivity decreases
Solution Approach 1:
The system performs preliminary assessments of document images using machine learning models to identify obvious issues before conducting detailed image processing. This preliminary action filters out documents that clearly fail to meet requirements, avoiding unnecessary detailed processing and reducing overall verification time while maintaining accuracy for documents that require full inspection.
Solution Approach 2:
The verification process is segmented into multiple stages: initial document detection, preliminary assessment using ML models, and detailed image processing only for documents that pass preliminary checks. This segmentation allows the system to apply different levels of processing intensity based on document quality, improving efficiency without sacrificing verification accuracy.
2Measurement precision
If detailed image processing is performed to determine whether a document meets institutional requirements, then measurement precision is improved, but productivity decreases
Solution Approach 1:
The system performs preliminary assessments of document images using machine learning models to identify obvious issues before conducting detailed image processing. This preliminary action filters out documents that clearly fail to meet requirements, avoiding unnecessary detailed processing and reducing overall verification time while maintaining accuracy for documents that require full inspection.
Solution Approach 2:
The verification process is segmented into multiple stages: initial document detection, preliminary assessment using ML models, and detailed image processing only for documents that pass preliminary checks. This segmentation allows the system to apply different levels of processing intensity based on document quality, improving efficiency without sacrificing verification accuracy.
3Ease of operation
If real-time remediation opportunities are provided to users, then ease of operation is improved, but device complexity increases
Solution Approach 1:
The system provides real-time feedback to users about document verification status and limit policy compliance. When documents or requests violate limits, the system immediately notifies users and offers remediation opportunities, such as modifying service requests or scheduling future deliveries. This feedback loop enables users to correct issues before extensive processing occurs, improving ease of operation while the underlying complexity is managed through automated ML-based assessment.
Data Source
AI summary
A computer implemented method, system, and non-transitory computer-readable device for a digital document verification process. In some embodiments, a split deposit tool may be provided in a mobile banking application to allow a customer to split a deposit and schedule deposit dates for various portions of the split deposit. In some embodiments, the split deposit tool may be provided in response to a determination that an amount of a check in an image exceeds a remaining remote deposit limit. In some embodiments, the amount may be determined using real-time optical character recognition (OCR), for example, active OCR performed on a live stream of check imagery.


