AI-based tools for better policing on the anvil

DUK training for cops to leverage technology and data

Developing Artificial Intelligence-driven software covering five critical areas, which will help police tackle better the changing contours of crime in an increasingly digital era, is under way with the assistance of the Kerala University of Digital Sciences, Innovation and Technology, also known as Digital University Kerala (DUK).

The initiative is part of the ongoing 45-day short-term training course based on ‘AI for All’ being held for 15 select cops, including 13 newly inducted civil police officers with technology backgrounds, as part of DUK’s capacity building initiative for government officials. The training primarily covered the application of AI, machine learning, image processing, and data analytics in policing besides an international workshop on the fast-evolving Brain Computer Interface facilitating brain reading of sorts.

“We are planning to give similar training to various government departments as part of the soon-to-be-set up Centre for Smart Governance at DUK to help them leverage technology and integrate data,” said Saji Gopinath, DUK Vice Chancellor.

Five projects were selected as part of the ongoing training for technology-aided policing - an investigation assistant chatbot, anti-social behaviour prediction on social media, crime hotspots prediction, culprits prediction using modus operandi, and face attribute manipulation. The course was designed by DUK Dean (Academics) Elizabeth Sherly.

“We asked all the 15 participants to come with the titles for 10 projects and then to submit an abstract of three of the most beneficial projects. It was further whittled down to five projects and the participants were split into teams of three each to develop separate AI-driven software for each project. They are set to present an initial programme module for the projects at the valedictory function of the training later this month,” said Malu G, Research Officer, DUK.

The investigation assistant chatbot is meant to lend assistance to an investigation officer by answering his/her multitude of queries right from the registration of FIR through to the course of the probe. The software for anti-social behaviour prediction on social media seeks to predict and verify fake, abusive, and hateful social media content by adopting deep learning-based analysis of social media behaviour.

The objective of the crime hotspot prediction is primarily to analyse existing legacy electronic data for trendsetting, crime analysis, disaster and traffic management thus facilitating better deployment of the police force. It also proposes linking databases across the country for big data analysis like predictive policing methods.

The culprit prediction programme aims at addressing the challenge of identifying the crimes committed by the same individual or groups by drawing parallels between crimes based on the modus operandi from the maze of criminal records with technological assistance.

The face attribute manipulation proposes to detect potential facial manipulations that a criminal may adopt to slip through the police network.