By cryptoadmin 
Earlier this year, Noida-based smart security and surveillance solutions provider Vehant Technology, secured Delhi Police as its client. The law enforcement agency has installed or is in the process of installing, 535 Automatic Number Plate Recognition (ANPR) software at strategic locations across the city.
The AI-powered ANPR system’s integration with police station control rooms enhances traffic surveillance by efficiently analysing video streams to recognise and capture vehicle number plates.
“We have covered over 2,000 lanes with our plate reading cameras, which are either deployed or are being deployed which essentially track vehicle movement in the city with regards to stolen suspected vehicles,” Kapil Bardeja, CEO and co-founder of Vehant Technologies, told AIM.
Worldwide, computer vision algorithms are becoming an indispensable tool for law enforcement agencies. Computer vision algorithms analyse video feeds in real-time, identifying and tracking people, vehicles, and helping law enforcement agencies to enhance public safety, investigate incidents, and respond promptly to potential security threats.
( Vehant co-founders Kapil Bardeja and Anoop G Prabhu)
AI for traffic enforcement
For example, Gurugram police have installed around 300 state-of-the-art cameras in the city for traffic enforcement, which detect violations such as speeding, no helmet, tripling, etc., and Vehant Technologies, incubated at IIT Delhi in 2005, is enabling this transformation.
The company leverages image classification, object detection, object tracking and image retrieval algorithms to improve the accuracy in the search and retrieval of objects of reference and expedites the decision-making capability of their clients.
“The models generally available are trained on generic data sets. The ones we use are application-oriented, and we customise them by training them for specific problem-solving. We run the models on edge devices. We take ideas from the literature and develop in-house solutions with performance at par with the state-of-the-art,” Anoop G Prabhu, co-founder & CTO at Vehant Technologies, told AIM.
“Our cameras are spread across 250 different lanes in Gurugram. Our system detects violations; let’s take, for example, someone riding without a helmet. It captures a snapshot, reads the number plate, and provides an image of the number plate. This information is bundled into a challan form and presented to the user for analysis in real-time with the help of AI,” Bardeja told AIM.
However, a manual intervention is involved as law enforcement agencies are mandated by law to manually verify if the challans are accurate and whether they should be subject to a fine.
“Once verified, it gets integrated with the NIC database of Sarthi and Vahaan where the information is fed and an auto-challan is generated. So this is a completely integrated seamless system,” Bardeja said.
Currently, around 2,000 challans are validated on a daily basis. However, Bardeja said that the number of challans generated by AI is way higher. Vehant’s number plate scanning technology is also deployed in 50 Smart Cities across India and the system manages close to 1 billion transactions a year.
AI-powered X-Ray baggage scanning
Vehant is the only Indian company approved by the Bureau of Civil Aviation Security to deploy dual-view X-ray scanners in Indian airports. So far, Vedant’s scanners have been deployed in airports at 70 airports in cities, including Chennai, Kolkata, Trichy, Pune, Udaipur, Varanasi, and Jodhpur, among others.
This year, Vehant also bagged a substantial order worth Rs. 37 crores from the Airports Authority of India (AAI) and Mumbai Airport to deploy its AI-powered dual-view X-ray scanners in Mumbai Airport, which is one of the busiest airports in the world.
The object detection algorithm, integrated into the devices, auto-detects contraband items such as guns, gun parts, knives, and scissors among other things.
“We use AI for detecting threats and restricted objects. The computer collects the information from the machine and identifies the objects using identification algorithms. We use AI-based multi-modal solutions for object and contraband detection,” Anoop G Prabhu, co-founder & CTO of Vehant Technologies, told AIM.
( Dual view X-Ray scanners)
The state-of-the-art machines are powered by AI algorithms, which automatically stops the conveyor immediately when it detects any contraband objects and alerts the operator of critical threats. This results in accurate and reliable detection for security operators with low false alarm rates.
“These are high-end technologies, and even though there are other Indian companies developing single view scanners, in the aviation market, only dual view scanners are sold, and we are the only Indian company which has managed to develop the technology,” Bardeja said.
Vehant’s ‘Make in India’ journey
Previously, most of the security and surveillance technology was imported from other nations. However, after the parliament attack in 2001, the government of India approached IIT Delhi for indigenously developed and under-vehicle scanning systems. Previously, such systems were imported from the UK and cost around INR 1 crore. This kickstarted Vedant Technology’s journey.
Recently, the company deployed its under-vehicle scanning systems at the new parliament building in New Delhi and Telangana secretariat building. However, back in 2005, when the company started out, according to Bardeja, the whole technology stack was not ready.
“Back then the computer resolutions were not that good and the hardware infrastructure, including compute power and GPU accelerators, was not as advanced as it is today. Running algorithms for specific use cases required substantial computational resources, making it economically impractical. Customers were hesitant to invest significant funds for use cases that demanded extensive computational resources,” Bardeja said.
However, in the last few years, significant progress has been made. Today, we have high resolution cameras and the compute cost has also come down significantly.
Even software algorithms, according to Bardeja, have progressed a lot over the years. The cost of bandwidth has also drastically reduced over the years in the country.
“Now, we can feed the data from the camera to the cloud at a relative low cost. I believe the amalgamation of these factors has allowed us to provide a reasonably accurate solution using our proprietary algorithm at a cost-effective price.”
The post This Indian Company is Harnessing AI for Traffic Regulation appeared first on Analytics India Magazine.