Bharat ABIS can detect duplicate Aadhaar IDs using an open-source biometric identification system developed by researchers associated with UIDAI.
The technology combines fingerprint, iris and facial data to search for duplicate identities at very large scale.
The system was evaluated using a de-identified gallery of 220 million Aadhaar records, randomly sampled from a database containing around 1.55 billion records, according to the research paper.
Bharat ABIS Can Detect Duplicate Aadhaar IDs
Bharat ABIS stands for Automated Biometric Identification System.
It is designed as a multimodal platform that processes fingerprint, iris and face information together, creating a combined biometric template of about 13.5 KB for each person.
The combined template can then be compared with existing records to identify whether a person’s biometric identity may already be present in the database.
The system also uses a scheduler to distribute search workloads across multiple servers.
How the New Biometric System Works
The research team included six UIDAI researchers, a researcher from IIIT Hyderabad and biometrics researcher Anil Jain of Michigan State University.
Their study, titled “Towards Billion-scale Multi-modal Biometric Search”, examines how biometric searches can be performed across extremely large identity databases.
The researchers tested Bharat ABIS against a 220-million-record Aadhaar gallery. The data used for the evaluation was de-identified, while the system itself was trained using smaller datasets.
The system can process fingerprint, face and iris information through separate stages including preprocessing, quality assessment and presentation attack detection before combining the biometric information for large-scale searches.
Bharat ABIS Reports Low Error Rates
The study reported an FNIR of 0.3% at an FPIR of 0.5% when tested with adult records in the 220-million-record evaluation.
On a 20-million-record gallery, the researchers also compared the system with three commercial biometric identification systems and reported comparable or better performance.
For large-scale processing, the researchers reported that Bharat ABIS could perform about 100 searches per second against a 40-million-record gallery using a single server equipped with eight Nvidia H100 GPUs and 2TB of RAM.
These results show the system’s research potential, but they do not by themselves establish that Bharat ABIS is already ready for full deployment across India’s entire Aadhaar database.
Why Duplicate Detection Matters
UIDAI’s Aadhaar-generation process already includes demographic and biometric de-duplication.
UIDAI says biometric de-duplication compares an individual’s biometrics against existing records to identify possible duplicates before an Aadhaar number is issued.
The new research focuses on making such large-scale biometric searching more efficient and accessible.
Bharat ABIS uses an open-source architecture, which the researchers say could also make similar technology more accessible to countries that cannot afford proprietary biometric identification systems.
However, further testing at larger scale is still important.
Performance measured on research datasets and large but limited galleries cannot automatically be treated as proof of operational performance across a database containing more than a billion identities.
