Officials use AI analytics to keep Bihar panchayat polls free and fair

The aim is to ensure that every vote in the electronic voting machines (EVMs) is error-free and not manipulated during counting.

The video analytics system used for the panchayat elections has been provided by a Gurugram-based artificial intelligence (AI) startup called Staqu. The system is based on its proprietary video analytics solution called JARVIS, which uses OCR and text recognition to identify the candidates and count votes directly from the camera feed by monitoring the EVM screens.

After the EVM data is fed into the systems, it is tallied with the data acquired by performing analytics of the footage from the CCTVs to ensure there are no discrepancies.

The data generated from the video analytics solution is hosted on the State Election Commission’s servers where the data going to its website and data generated by Staqu’s solution are matched.

If any difference in the number of votes is detected, the commission is alerted immediately.

This is the first time that video analytics is being used during vote counting in any election in India.

The State Election Commission has also deployed a fingerprint recognition system to verify the identity of the voters.

The panchayat elections in the state will be conducted in 11 phases from 24 September to 12 December.

“Panchayat elections are always a huge election as we have a large number of polling booths and candidates. Given the number of candidates and number of panchayats we receive in each election, we felt it best to automate the vote-counting verification process,” said Deepak Prasad, Bihar state election commissioner.

“We are happy to associate with Staqu’s artificial intelligence-driven JARVIS technology, which can assist in the process and ensure that it is error-free and the elections proceed in a fair manner,” he added.

Unlike state and national-level elections, most panchayat elections don’t use VVPAT (voter verifiable paper audit trail) for verification of votes.

“We have requested district magistrates to ensure that a camera connected to JARVIS is placed on the same table where the EVM is kept and is facing and detecting the EVM machine. Then, we ensure that it is detecting the text in the machine and the OCR is identifying the characters and their meaning,” said Atul Rai, chief executive officer and co-founder at Staqu.

The whole idea was to address any scope for discrepancy during the counting without having to interfere with the EVM, he added.

Rai claims the accuracy of their OCR is 99.7%.

Staqu has been one of the leading providers of AI, face recognition and video analytics to law enforcement agencies in India.

In 2019, the Uttar Pradesh government had deployed JARVIS for its face recognition and intelligent monitoring capabilities in 70 prisons to monitor prisoners.

Government agencies have also started leveraging emerging technologies for governance and law enforcement.

The use of drones by police for crowd control during the Kumbh Mela and Chhath Puja, for instance, is now a common feature.

During the 2020 protests against the Citizenship Amendment Act, the Delhi Police had deployed facial recognition software to identify protesters indulging in violence. In January, the UP government announced it will use AI-enabled cameras to read facial expressions and help women in distress.