TrustStamp: Enhancing the Accuracy of Identity Verification with Amazon Textract
Information Technology & Services SportsBusiness Impacts
88%
extraction accuracy for processed documents
5.03s
latency reduction for quicker turnaround time
Customer Key Facts
- Country : Atlanta, Georgia, United States.
- Industry : Software Development
Problem Context
Trust Stamp, a leading provider of AI-based identity services across various industries, was planning to shift from their current solution for extracting identity information. The goal of this pivot was to further improve the accuracy of the information captured while reducing the high maintenance costs associated with their current solution
Challenges
- Had issues w.r.t the accuracy and processing turnaround time of the extracted information
- Maintaining their existing solution that was labor-intensive, leading to higher costs
Technologies Used
Amazon Textract (AnalyzeID API)
AWS Lambda
Amazon Simple Queue Service (SQS)
Amazon S3
Amazon Aurora RDS
Enabling identity information extraction from different ID Documents
Solution
- Quantiphi developed a custom solution enabling the extraction of identity information from different ID Documents issued by the US Government - driver's licenses, state IDs, and passports.
- Quantiphi used a sample size of 171 documents and categorized them into three groups according to their visual appearance and image quality. Additionally, we undertook a comprehensive comparative analysis of the output with the existing solution and generated reports for each type of ID document.
Results
Successfully demonstrated our expertise in document processing and text extraction, resulting in quicker processing times with improved accuracy of 16.5% from the existing solution.
"Quantiphi assisted in demonstrating the possibilities of a new Amazon technology and testing the value of the solution. Quantiphi's team was extremely professional. They came prepared for every meeting and completed all of their deliverables on time. They interfaced well with our team and met all of the program objectives."
Scott Francis, Chief Technology Officer
