|
|
|
You can get e-magazine links on WhatsApp. Click here
|
|
|
|
|
|
AI expands role of CCTV in retail and F&B sectors
|
|
Wednesday, 30 September, 2026, 08 : 00 AM [IST]
|
|
Nandita Vijayasimha, Bengaluru
|
In retail and F&B sectors, artificial intelligence (AI) is expanding the role of CCTV by turning video feeds into actionable business intelligence. This helps improve efficiency, enhance customer experiences, reduce losses, and make smarter decisions using data already being captured through their CCTV systems.
Kunal Kislay, Co- Founder & CEO, BLUE AI (KGraph AI Solutions), said a quick service kitchen produces more operational signals in a single shift than its POS (point of sale) records in a month. Yet almost none of it is used. The cameras are running and the footage is stored, but it is usually opened only after something has already gone wrong. That is the legacy of CCTV built primarily as a security system. It answers one question, retrospectively, for one department.
The gap it leaves is widest in F&B, because so much of what determines margin happens in physical space. A POS records that an order was placed at 1:05 PM. It cannot say whether a queue had formed, whether the handover was late, whether the table was cleared, or whether the void rung at 9 PM had a customer standing in front of it. A WMS records that goods were received. Only video can show how they were handled. A processing line records throughput, but not whether hygiene and PPE protocols were followed during the night shift, he added.
Operators fill that gap with audits, checklists and area managers driving between sites - sampled, lagging and largely self-reported. Two hours of a 720-hour month get inspected and the rest is extrapolated. With food safety enforcement tightening across Indian states and quick-commerce compressing service expectations, that sampling error is getting expensive, Kislay told FnB News in an email.
The obstacle has been that video does not behave like enterprise data. Thousands of cameras generate continuous streams, and the bandwidth, storage and compute cost of moving and analysing them at fleet scale has kept deployments small and single-purpose. Traditional computer vision compounds it: each camera must be told in advance what to look for, so every new operational question becomes another model and another deployment cycle, he said.
We at BLUE have approached this as an infrastructure problem rather than simply a detection problem. The AI-native platform is designed to understand video once and make it possible to query that information many times. Rather than processing every frame as equally important, the semantic video technology interprets what is happening and stores a structured representation of it. This can reduce the bandwidth, storage and compute required to run video at fleet scale, while making footage more accessible to AI systems rather than limiting it to a person manually scrubbing through a timeline.
The platform is currently live across roughly 500 sites and 4,000 cameras, including deployments with Britannia, Haldiram’s and CPRL, the McDonald’s licensee for North and East India, across food processing, warehousing and QSR environments. These deployments do not require existing infrastructure to be replaced. The intelligence layer sits on top of installed CCTV and NVR systems, he said.
Enterprises already run ERP for transactions, CRM for customers and IoT for machines. Video is the last large operational dataset still outside that stack, and the sensors to bring it in are already on the wall, said Kislay.
|
|
|
|
|
|
|
|
|
|