The government of Maharashtra is cleaning up voter lists using IT tools, to minimize duplications in voter ID cards and correct errors.
Face-recognition algorithm and geographic information system (GIS) technology are examples of the information technologies being used to address errors in names, postal addresses and attend other electoral issues.
Mr Rajesh Aggarwal, principal secretary at the department of Information technology, government of Maharashtra, said: “We shall complete the modalities and framework within this May 2014 and finalise the dates for commencement of the drive."
Maharashtra is the second most populous State after Uttar Pradesh. It has approximately 8.06 core voters comprising 288 assembly constituencies. Maharashtra will go to polls in five months.
IT expert and head of NIC at Andhra Pradesh – the official IT agency under government of India – Mr BV Sharma said this is a first.
"As per my knowledge, the usage of face-recognition algorithms for correction of voters' lists, if put into practice, is the first time in India by a state government," he said. Currently, BSNL, a government of India undertaking uses face-recognition algorithms for its telephone directory, he said.
Maharastra has many credits to it for IT-savvy electoral reforms. One such initiative, the Election Management Project, is aimed at integration between the websites of the state election commission, local bodies/urban local
Before e-Yadi, electoral roll preparation involved manual cutting, pasting and copying. A strict monitoring system could not be established and, as a result, the electoral roll preparation process was full of loopholes.
Another is the introduction of GIS technology to demarcate election wards using Google maps. To make elections hassle-free for all stakeholders, the State Election Commission has implemented it. The maps show the boundaries of the selected wards and the polling booths during polls. The current scenario enables easy and direct access to citizens given by the commission as to where and when to vote.
Recognition algorithms can be divided into two main approaches: geometric, which looks at distinguishing features, or photometric, which is a statistical approach that distills an image into values and compares the values with templates to eliminate variances.
According to Mr BV Sharma, face-recognition algorithms are available (with Microsoft and Oracle) and they are suitable for the use of recognising similar images and identifying similar names and grouping for correction in English. However, it is not known whether these algorithms can be used for regional language (Marathi).