Gujarat Police Seeks AI System to Connect 80,000 CCTV Cameras

Gujarat Police is seeking technology to integrate approximately 80,000 CCTV cameras across the state into a unified video management and analytics ecosystem through the Gujarat Police Innovation Hackathon 2026.

The challenge offers a total prize pool of ₹37 lakh and is scheduled for September 10–11, according to the official Sentinel Gujarat website.

The project aims to address a fragmented surveillance infrastructure. The government says 26 different departments currently operate independent CCTV systems using different vendors, video management systems (VMS), storage arrangements, camera types and feed-sharing protocols. Some systems retain footage for seven days, while others retain it for 15 days or more.

The proposed system would allow these separate networks to work through a common integration layer while, under some of the proposed models, allowing existing departmental systems to continue operating independently.

The government says the architecture should be secure, scalable, interoperable, and designed to use existing infrastructure to the maximum practical extent.

The system is also expected to connect CCTV analytics with databases including VAHAN, SARTHI, eGujCop (Gujarat Police's CCTNS platform), the Automated Fingerprint Identification System, and the National Automated Fingerprint Identification System. These databases contain records relating to arrested persons, stolen vehicles, wanted criminals, missing persons, unidentified dead bodies, and fingerprint data, according to the government.

For the technical evaluation, participants will receive access to approximately 50 geographically distributed cameras. They must onboard the feeds into one platform, track a designated vehicle using its registration number, reconstruct its timestamped, and location-based movement history and generate real-time alerts from a representative watchlist database.

The requirements include automatic number plate recognition, person and vehicle tracking, object detection, and event analytics. Facial recognition systems are also listed among the proposed analytics capabilities, while the government requires participants to demonstrate cybersecurity, scalability, and interoperability.

Participants can choose among four reference models or submit a hybrid or customised architecture. The models range from a central CCTV registry and Geographic Information System (GIS) foundation to unified viewing, VMS federation, and a fully centralised VMS and artificial intelligence platform.

The government will evaluate successful feed integration, the working platform, analytics output, architecture, security and readiness to scale toward approximately 80,000 cameras. It may also consider privacy protection, auditability, role-based access controls, edge processing, and bandwidth optimisation.

The official website says registration opened on August 4 and closes on September 7. Shortlisting is scheduled for September 7, with the event on September 10–11 and results on September 11.