Skip to main content Skip to footer
  • "com.cts.aem.core.models.NavigationItem@555a73fb" Careers
  • "com.cts.aem.core.models.NavigationItem@5fc969d6" News
  • "com.cts.aem.core.models.NavigationItem@4a7e66ac" Events
  • "com.cts.aem.core.models.NavigationItem@efcf58a" Investors
Case study

The challenge

Industries in India have been disrupted recently by the 2019 Aadhaar Act privacy amendment. Aadhaar numbers are unique identification numbers issued to every resident of India by the government. This new law requires all organizations in India to comply with provisions to mask each individual’s Aadhaar number for security purposes.

Seeking to comply with this new mandate—and avoid a stiff penalty—one large Indian insurance company enlisted Cognizant to cloak the numbers using an automated process driven by artificial intelligence (AI). As the law prohibits intermediate storage of data, the company wanted an efficient masking technique without disrupting its existing processes.

Our approach 

Cognizant created an AI-enabled smudging solution for the insurer’s historical heterogeneous data, storage devices and future real-time Aadhaar image processing.

This process was carried out by taking the image input and providing a masked image as the output. A cognitive service-based machine learning model, running on the Microsoft Azure platform, masks the images in the background.

Our solution facilitates identification of long or short format Aadhaar cards and masks bar codes, QR codes and identification numbers. Azure Computer Vision technology identifies the Aadhaar numbers in the image. A machine learning model, based on Azure Cognitive Services and trained using sample data provided by the client, identifies QR codes and bar codes.

Faster, scalable, fully compliant protection

This automated, secure and maintenance-free solution is much faster than the insurer’s previous manual image masking technique while remaining fully compliant with process requirements. There is no risk of data leakage because the API does not store images during processing. The new system is also scalable to support large changes in workload and throughput.

99%

reduction in operational costs over manual masking

100%

image masking accuracy

24/7

card image system availability