[ad_1]
Drive through any Indian city today and it feels as if every corner, from traffic junctions and metro stations to markets, offices, schools and housing societies, is under surveillance. With over 84,000 CCTV cameras already installed across 100 smart cities and integrated into citywide command-and-control centres, India has built one of the most extensive urban surveillance infrastructures in the world. Yet, whenever an incident occurs, the familiar line resurfaces: “CCTV footage is being examined,” and by the time hours or days of video are manually reviewed, the trail often goes cold and the culprit is not discovered.
This paradox exposes a harsh truth: the problem is not the lack of cameras, but the way they are used. Traditional cameras, no matter how deployed, are essentially silent devices until a human steps in to view, interpret, and act on the footage. They record everything but understand nothing. In fact, cities have created vast collections of visual data that are incredibly rich in evidence, yet it is extremely slow and expensive to access them when needed at the time.
India’s surveillance work has been intensified by programs like the Smart Cities Mission and the rapidly growing CCTV and electronic security market. Government data shows that more than 84,000 cameras across 100 smart cities are already linked to integrated command and control centers (ICCCs) for public safety, traffic enforcement and incident monitoring. Industry analysis shows that India’s CCTV and video surveillance market is worth several billion dollars and growing strongly, indicating that investment in visual infrastructure will deepen in the coming years.
Yet, everyday policing experience often tells a different story. When a robbery, assault, or hit-and-run occurs, investigators typically stitch together footage from multiple public and private cameras, then manually sift through hours of recordings to identify a key moment. In many cities, a significant portion of footage requests cannot be fulfilled in time due to fragmented ownership, storage limitations and lack of manpower, leading to low resolution rates in categories such as snatching and street crime despite the presence of cameras.
The main limitation is architectural: traditional CCTV systems were designed as recording tools, not real-time decision systems. They capture and store data but rely on the human eye for detection, correlation and decisions. in practical terms:
- A camera can “see” a person brandishing a weapon, but it cannot itself raise an alarm.
- It can record the snatching incident, but cannot automatically mark the suspect’s face or clothing for tracking in other feeds.
This human-in-the-loop model is not scalable in cities that generate millions of hours of video every day. Even with ICCCs and monitoring rooms, a limited number of operators cannot meaningfully track hundreds or thousands of live feeds simultaneously. As a result, cameras become “silent witnesses”, useful after the fact, often in a limited way, but rarely deterrent in real time.
The recent case on Purvanchal Expressway in Uttar Pradesh clearly shows the privacy risks of this human-dependent model, where uncontrolled access turns surveillance into a weapon against citizens. A toll plaza assistant manager allegedly took advantage of high-resolution CCTV feeds by zooming into parked vehicles to secretly record the most private moments of commuters – including the moments of a newlywed couple – then used the footage to demand extortion payments and even leaked it online after one such incident took place. The incident, which sparked complaints to top officials, shows that handing over cameras to human operators without monitoring exposes everyday people to harassment, blackmail and dignity violations that are worse than any monitoring gap in crime detection.
Artificial Intelligence (AI) changes this equation by adding a real-time analytical brain to the existing visualization infrastructure. The AI-enabled Setup platform can analyze hundreds of thousands of frames per second with sub-second latency, transforming raw footage into searchable, actionable intelligence. Investigators can perform text-based video searches such as “man in red T-shirt” or “person with green bag” or run facial recognition and object detection on distributed camera networks within minutes instead of manually scrubbing through hours of video.
Importantly, AI also enables policy-driven controls that reduce privacy invasions enabled by uncontrolled human visualization, as seen on the Purvanchal Expressway. Features like role-based access restrictions, automated audit logs of all footage views and exports, automatic masking of sensitive areas or faces in non-threatening scenarios, and real-time alerts for unusual zooming or prolonged investigations can enforce privacy-by-design without relying on personal ethics. In this framework, the system detects and prevents potential abuse by the operators themselves, protecting citizens from insider threats while empowering proactive crime prevention.
Across India, AI-enabled CCTV is already demonstrating tangible law-enforcement benefits. AI video analytics and facial recognition are being used by police forces to identify suspects in major robbery cases, trace escape routes, and recreate crime timelines in a fraction of the time previously required. During high-security incidents, AI systems monitor thousands of live feeds to detect suspicious behavior, unattended objects, crowding, or perimeter breaches and generate real-time alerts for security teams. In some cities and critical facilities, AI-powered cameras now automatically detect loitering, vandalism, or the presence of weapons, enabling proactive intervention rather than a delayed response.
The lesson for India’s surveillance strategy is clear: adding more lenses without adding intelligence will only deepen the collection problem. The country is already on a strong growth trajectory for AI-enabled CCTV, with the AI CCTV market in India valued at around $827 million in 2023 and projected to grow more than four times to around $3.67 billion by 2030. This reflects a strategic shift from simply installing cameras to deploying systems that can detect, correlate and alert in real time.
For policymakers, city administrators, and enterprises, the imperative now is to consider AI not as an add-on but as the core of any new surveillance investments. Cameras must evolve from passive recorders to active partners in security systems that can identify threats, support officers on the ground and help prevent heinous incidents before they escalate. Otherwise, the risk is clear: India will continue to spend heavily on technology that monitors everything, yet protects very little in a meaningful way. In smart cities, there is a demand for eyes with smart cameras that can not only see, but also understand and help in solving crimes.
This article is written by Atul Rai, CEO and Co-Founder of Stack Technologies.
[ad_2]
