From Pizza AI to Operational Intelligence: How Sbarro Turned Existing Cameras Into a Business Intelligence Platform
How Sbarro and Camcloud combined existing cameras, custom AI, and proprietary KPIs to create a scalable way to measure and improve restaurant operations.
From Video Footage to Business Intelligence
What if the cameras already installed in your stores could do more than record surveillance footage? That was the question Sbarro set out to answer with Camcloud.
For a global multi-location restaurant operation, achieving the highest standards across its locations is an ongoing business priority. Sbarro’s operations team needed a scalable way to measure food display quality, improve operational consistency, supplement manual audits with objective measurement, increase accountability across locations, and ultimately connect store execution with broader business performance.
Traditional video surveillance and NVRs or VMS systems can provide the image and video footage, but footage alone doesn’t provide an operational measurement and is hard to scale.
Sbarro wanted to answer a different question:
Could the cloud-connected cameras already deployed throughout its restaurants combined with cloud-base AI become a scalable source of operational data, specifically the quality of its food displays?
To answer this question, Sbarro and Camcloud developed a custom computer vision (AI) solution designed specifically around Sbarro’s restaurant environments and operational requirements. The cameras remained in place. What changed was what could be done with the video and image data they captured.
What started as an experimental AI initiative, called “Pizza AI,” became a much larger operational initiative: using existing store cameras and custom-trained AI models to continuously measure food display quality and turn those observations into actionable business intelligence.
Today, what started as an experimental pilot, is a solution that’s now deployed across Sbarro’s corporate restaurant network and is being promoted across its global franchise network.
Why Custom AI Matters
At the foundation of the solution is the existing camera infrastructure. Leveraging already deployed IP cameras at all its stores, Camcloud’s machine learning team worked directly with Sbarro to train and refine custom computer vision models using real-world imagery from live restaurant environments.
The models were developed specifically to interpret food display conditions and collect the measurements that mattered to Sbarro. The process can be thought of as a series of steps:
Instead of requiring new cameras or deploying an expensive purpose-built hardware system, the solution uses video and images from cameras already deployed in Sbarro locations. The custom models analyze the footage and convert what they see into measurable quality signals.
Those raw measurements are then incorporated into Sbarro’s proprietary KPIs and management systems, creating an AI-powered scorecard for store performance.
Restaurant environments are dynamic. For that reason, the models needed to be trained around Sbarro’s specific environments and workflows.
To that end, Camcloud’s machine learning team worked with Sbarro to iteratively train and refine the models using real-world operational imagery. As more training data and operational feedback became available, the models became increasingly resilient to challenges such as:
- Obstructions
- Challenging camera angles
- Lighting and image-quality variations
- Motion and active food-service conditions
- Other operational edge cases
One of the key differences between simply applying an off-the-shelf analytics to a video feed and building an AI solution around a specific business problem is that the models continue to evolve as additional training data and business feedback are incorporated.
From AI Experiment to Operational Intelligence
The computer vision models provide the observations. The real value emerged when Sbarro integrated the AI-generated data into its proprietary store-performance KPI framework, with executive management buy-in and continuous improvement processes supporting the deployment, which transformed the initiative from an interesting computer vision experiment into an operational intelligence platform.
AI doesn’t exist in isolation. It generates signals from the physical environment. Sbarro then applies its own business rules and KPIs to those signals, turning them into information that managers can use to understand performance across locations.
The rollout happened in stages: the initial AI experiment tested whether existing store cameras could generate meaningful operational insights. After the models were trained and validated using real store data, the solution was deployed across a small number of pilot locations, which grew as confidence increased in the results.. Following successful pilot results, the solution expanded across Sbarro’s company-owned restaurant system. Now, the initiative is being promoted across Sbarro’s global franchise network.
The resulting system gives Sbarro a way to objectively measure food display performance across its restaurant system. Weekly operational reports provide managers and leadership teams with visibility into performance, while the resulting insights can feed into ongoing operational improvement.
The Camcloud Advantage
A camera doesn’t have to be limited to security. The same infrastructure that provides visibility into a physical location can also become a source of operational data when combined with the right AI models, business rules, and analytics framework.
Camcloud is designed to work with existing IP camera infrastructure, providing cloud video management and AI-powered analytics without requiring organizations to replace their deployed cameras or build complex onsite infrastructure.
For many businesses, the move to the cloud starts with a simple problem. They may be looking to replace an aging NVR system, make their cameras easier to access, or move video management out of the local environment. But once that foundation is in place, the possibilities can grow.
A business might start by moving its existing cameras to the cloud. Then, it may want to use those same cameras for AI-powered analysis. From there, it could add tools for monitoring and surveillance outside of business hours, or look for new ways to use video to understand what is happening across its locations.
Instead of having to replace an entire system every time a new need comes up, businesses can build on the infrastructure they already have and add new capabilities as their needs evolve.
Ready to explore what your existing cameras could do?
Camcloud helps businesses connect, manage, and access video from virtually any IP camera while providing the cloud infrastructure and AI capabilities needed to build smarter security and operational applications.
No Comments