Artificial Intelligence (AI) is fast evolving as a go-to technology for bringing the best to humans. Gurugram-based startup Staqu is an example in India, and an AI app developed by the firm is helping law enforcement officers across the country tackle crime and check its impact.
The app, named Artificial Intelligence Based Human Efface Detection (ABHED) is helping law enforcers digitise criminal records and retrieve the information in real-time through image analysis, language and text independent proprietary speaker identification engine, facial recognition and text processing, including name entity recognition, sentiment analysis and summarisation APIs.
Till date, the startup has helped the police forces of Rajasthan, Punjab, Uttar Pradesh and Uttarakhand solve more than 1,100 cases. It also helps Indian Army to conduct aerial imaging analysis.
In an exclusive interview with CNBC-TV 18, Atul Rai, CEO and Co-founder of Staqu , explained the nitty-gritties of the application and the role of artificial intelligence in helping police forces nab criminals and fraudsters.
Edited excerpts of the interview: How does Staqu operate and help the police curb and solve crime? Can you give a few examples of cases solved by Staqu?
At Staqu, we had forayed into assisting police forces with the launch of our AI-powered app, ABHED. The app helped the task forces digitise criminal records and retrieve the information in real-time through image, text, or speech-based search. Starting from a pilot project in Rajasthan, the app has now reached Punjab, Uttarakhand and Uttar Pradesh.
Not stopping there, Staqu also launched a smart-glass solution for the task forces. The AI software, when integrated with the smart glasses, allowed law enforcement personnel to scan through crowds of people and locate the criminals camouflaging among large groups.
The company is coming up with innovations meant to assist the police forces in solving crimes. This include GAIT Analysis, which takes the real-time footage and identify criminals through their mannerisms, the way they walk etc. With GANG Analysis, we are able to help police nab gangs of criminals who may be active in different regions. This is coupled with Hotspot and Geo-Fencing Analysis that scan through the historical data and nearby places from a crime scene to generate leads for investigators.
At present, we carry a database of more than one million criminals in India and have helped police solve 1,100+ high-risk cases.
How does this app help investigators track missing children and spot crime scenes/dead bodies?
Staqu has been enabling the search for missing people through AI-powered facial recognition. In all the states, whenever a missing person report comes to the police station, the name and photo (which is provided by the parents or relatives) are uploaded to the app. In case any police personnel find any suspicious children or a dead body, they can click the picture and our app matches it to the database.
We are working to make the system more citizen-centric so that if anybody found any missing children or suspicious person on the road, they could click the photo which could be matched and searched in all the states and the report could be sent to the local police station from where the match was found.
The solution is currently live with three state-level police departments– Rajasthan, Uttarakhand, and Uttar Pradesh, with more integrations with other state-level police departments in the pipeline. Of course, in the next phase, the live streams from public cameras will be integrated into the system.
The promoters of Staqu -- Pankaj, Atul, Chetan and Anurag.
Can you explain what is speech detection and accuracy and how does it function with the help of this app?
The solutions have the capability to enable search by matching any biometric feature – face, fingerprints or voice. Voice samples prove to be very handy in identifying ransom callers or cybercriminals, who tend to be repeat offenders. Our company has been in top five 97 percent accuracy in matching voice samples and our research also got accepted in
Interspeech conference, one of the most impactful AI conference in audio and speech domain. The technology utilises the latest deep learning-based AI techniques to extract features from any speech samples which are language and text independent. That means whether you speak Tamil, Hindi or French, the technology will identify you uniquely. What is the future of smart glasses in India? Does the AI-powered glasses really help law enforcement officials identify criminals?
To understand the use of AI-powered smart glass, let’s take an example of an election rally or an important delegation. Security personnel are at high-alert during such events, and yet don’t have any mechanism to scan the crowd and get information regarding the background of different people present there. A criminal in disguise may cross a police check-point unnoticed. However, our technology solution addresses these limitations.
The company utilises the inbuilt camera of smart glasses to take inputs of the faces, for instance the iris of the human eye. This information is then streamed through Wi-Fi or 4G at the security hub. The device is already loaded with data from the security records containing details of the criminals.
Hence, post the streaming of the live video, the device checks the data against the available records in real time. If in case the technology recognises a suspect, the security personnel are alerted through the same smart glass. The alerts appear as image cast, projected on the screen of the glass, right before the eyes of security personnel.
The solution is going to have a tremendous effect on the functioning of law enforcement and crime control units. Different security personnel and police forces can utilise this solution, be it to safeguard the government or VIPs, ensure the smooth procession of election rallies and other such similar services.
ABHED is an innovative mobile phone-based criminal registration system that helps police forces easily nab criminals or search for missing people in the following ways:
How does ABHED function? Every police personnel will have its own login password to use the app for both, registering and search. With a mobile phone app, police personnel can automatically profile criminals or missing people, including biometric information like fingerprints, voice and face pictures. ABHED provides matching details from the criminal database within seconds, offers better accuracy than any other criminal recognition system. Criminal search can be performed through FIR, available face image or fingerprints. ABHED’s admin panel can be used to easily register or search for criminal or missing people. The bi-lingual admin panel (Hindi and English) can also be extended to cover other languages. Centralised web portal for different crime analysis. The system is secured by latest AI techniques. How do you look at the future of Artificial Intelligence in solving criminal cases?
Andrew NG once said AI is the new electricity. Going by the same notion, I believe that AI is a mechanism that can be used to extract meaningful information from any given big data. Crimes, especially in India, tend to be repetitive in nature, leaving a long trail of data, which can be utilised to reveal significant insights for the law enforcement agencies.
Can you tell us about your current tie-ups and future plans?
In India, different state-level police forces are utilising Staqu's AI-enabled solutions. For instance, in Rajasthan, we launched the pilot project of ABHED with Alwar police with the help of Rahul Prakash IPS, that digitised criminal and missing persons records and helped police forces retrieve this information in real time.
The pilot programme then reached Punjab, in the form of Punjab Artificial Intelligence System (PAIS), under the leadership of Nilabh Kishore, IG of the Organised Crime Control Unit of Punjab. Quite recently, we also collaborated with Uttarakhand state police, providing them with police artificial intelligence system. The solution is also live with the Uttar Pradesh police department as Trinetra, equipped with advancements like GANG and GAIT Analysis.
Staqu is also working with Dubai Police to implement an Artificial Intelligence-based predictive policing technology. We have been selected by the Dubai Police to establish predictive policing protocol.
How do you look at predictive policing in India? Do you see any future for it?
Back in 2017, violence cost India $1.19 trillion, or 9 percent of India’s total GDP that year. By adopting advanced technologies like digitalisation, artificial intelligence and big data analytics in the domain, we look at drastically reducing this number.There is certainly a bright future for predictive policing in India. With the growing levels of digitalisation in the nation, along with the advent of Smart Cities, we would have at our hands, humongous raw and unstructured data. With AI-powered big data analysis, we can unlock unprecedented efficiencies for the task forces, adequately equipping and empowering them with advanced technologies to ensure the smooth functioning of civil societies.