The solution

Virtusa developed an automated label verification tool leveraging AWS genAI services, including Amazon Bedrock and Amazon Textract. This genAI-powered solution automatically extracted crucial information from product label images, such as Product Description, Bioengineered Statement, Enjoy Through Date, and Lot Code—and compared it against a predetermined truth dataset to ensure consistency and accuracy.

The tool generated comprehensive reports detailing the matching rates and flagged discrepancies between the extracted and reference datasets. Additionally, it meticulously tracked performance metrics such as token usage, processing time, and an overview of image processing steps. These insights allowed for ongoing refinement and optimization of the verification process.

Virtusa implemented genAI models to boost the system's effectiveness as part of the solution. The client learned that the quality of results depended on the genAI models used—more advanced and latest models resulted in higher accuracy rates and better overall performance. With continuous improvements to the system, the tool achieved a matching rate of over 95% for non-complex labels, significantly reducing the error rate and manual intervention required for failed matches.

AWS services used:

Amazon Bedrock
Amazon Textract
AWS Lambda
Amazon API Gateway
Amazon DynamoDB
AWS Storage Gateway
Amazon S3
Amazon EventBridge
Amazon SQS
Amazon CloudFront
Amazon Route 53
Amazon VPC

Services

Industries

Insights

Careers

About