Problem/Objective

Getting the annual life certificate is a cumbersome task for pensioners as they have to travel long distances to a government establishment to give proof of liveness and to establish identity. This manual process, mandating physical presence of the old pensioners, is not very convenient for the citizens or the authorities, and there is often scope for errors. The Telengana government has scrapped the old system and replaced it with an app-based authentication. All that is needed is a smartphone and internet, and by uploading a freshly-taken selfie on the app, the pensioners can bypass the otherwise tedious process of visiting the local Treasury department to get the pension. The whole process takes just under one minute to find answers for the two crucial questions in authenticating a pensioner – Whether the pensioner in question is alive? Is she or he the legitimate pensioner?

Solution/Approach

Telangana state government has implemented the Realtime Digital Authentication of Identity (RTDAI) in authenticating pensioners. A smartphone is the only hardware requirement on the user end as fingerprints or iris images are not required. The beneficiaries have to take a photo and upload it to the exclusive app, which is offered as a part of the T App Folio – the umbrella app for Telangana's e-governance initiatives. The Pensioners Life Certificate Authentication through Selfie (PLCS) method deploys three levels of authentication: demographics (name, father’s name and address), photo and liveness. PLCS uses an artificial intelligence-based liveness check solution, big data and machine learning based demographic check solution, and deep learning-based image comparison solution. The AI, ML and deep learning solutions quickly check the details submitted by the user with the information piled up in public data systems.

Impact/Implementation

The AI tools can address challenges such as identifying old pictures in the data base from the latest photos of users as the system can detect the changes happened in faces over a period of time. Similarly with the names, the AI-enabled system can match names even with minor variation in spellings as it checks other parameters in the data, such as address or father's name. The success rate in authentication of pensioners is about 93 per cent, and it can be as high as 96-98 per cent as the system learns and improves over time.

Image from needpix.com