By Venkat Ramana, CEO, NthEye & Value Pitch
The Delhi Police plans to deploy 10,000 AI-enabled cameras under its Safe City Project, reflecting the growing role of technology in public safety. But as our cities become denser, public gatherings grow larger and security threats evolve faster, we need to ask a more fundamental question. Is adding more cameras enough to make our public spaces safer?
For decades, surveillance has largely been about recording what happened and returning to footage after an incident. That approach has served us well, but it is no longer enough. The next step in public safety is decision intelligence, in which AI translates real-time visual data into contextual, useful insights that support authorities in understanding situations and acting faster.
India does not simply need more cameras. It needs smarter intelligence layered over the surveillance infrastructure already in place.
The problem is no longer visibility
Indian cities generate enormous volumes of surveillance footage. Yet no command centre can expect human operators to continuously watch thousands of feeds and identify every abandoned object, unusual movement or emerging crowd surge.
The limitation is not the camera. It is the gap between seeing and understanding. AI-powered video analytics is closing that gap. Computer vision can analyse live feeds, identify anomalies and prioritise incidents requiring human attention. A person loitering near a restricted zone, an unattended bag, reverse crowd movement or a sudden rise in density can be flagged within seconds.
This fundamentally changes the role of surveillance. Cameras stop being passive witnesses and become part of an intelligence layer that helps authorities interpret events, prioritise threats and make faster, more informed decisions.
Public safety cannot wait for hindsight
AI’s value is particularly evident at large public gatherings, where millions can move through limited spaces within hours. By analysing crowd density and movement patterns in real time, AI can identify congestion or sudden surges early, allowing authorities to intervene before situations become dangerous.
Real-world deployments demonstrate this shift. In Kalyan-Dombivli, AI-powered crime heat maps identified incident hotspots, helping redesign patrol routes and improve coverage in high-risk zones. In Pimpri Chinchwad, video analytics detected unattended objects, intrusion attempts and perimeter breaches, while ANPR helped track suspect vehicles in high-alert areas. These examples show how real-time intelligence can improve crowd safety, crime prevention and operational efficiency.
The answer, therefore, is not to keep multiplying cameras. It is to make existing camera networks more intelligent, contextual and capable of supporting action when it matters.
From detecting faces to understanding behaviour
Facial recognition often dominates conversations around AI-assisted policing. But identity matching alone should never be mistaken for complete intelligence. A responsible system must understand context.
A face match may assist identity verification, but behavioural signs and situational evidence are as important before intelligence becomes actionable. This can reduce false positives and ensure technology supports expert judgement rather than replacing it.
The same principle applies to behavioural analytics. AI can detect loitering, perimeter breaches, abandoned objects or unusual movement. The final decision, however, must remain with trained law enforcement personnel. AI should sharpen human judgement, not automate authority.
From hours of footage to faster answers
In the case of traditional investigations, officers may have to study hours of footage from several cameras manually. Metadata provided by AI can make video searchable by time, location, object or movement pattern. Investigators can more quickly identify key sequences and reconstruct events more precisely.
Additionally, systems can generate a digital trail in chronological order as events unfold when issues are identified live. This may minimise inquiry time, improve administrative efficiency, and support subsequent investigations and legal proceedings.
For police forces managing growing volumes of digital evidence, this capability will become increasingly important.
The next smart city will need an intelligence layer
India has already invested significantly in surveillance infrastructure. The next priority should be making that infrastructure think smarter. By connecting video feeds, traffic systems and sensors, AI can give authorities a clearer operational picture and enable faster emergency response.
This intelligence must be built on accuracy, accountability, data governance and human oversight. Public safety threats are evolving too quickly for surveillance to remain reactive.
The future of public safety will not belong to the city with the most cameras but to the city that can transform every camera into timely, actionable intelligence. That shift from passive surveillance to proactive decision-making will define the next generation of safer, smarter cities.