Using ML Computer Vision to ensure safe and healthy restaurant dining

Making dining during COVID-19 safe and possible

Restaurants are one of the industries that has suffered the most operating hardship during the COVID-19 pandemic. Numerous shutdowns and reopenings have damaged consumer confidence in dining out. In an industry that already operates on razor-thin profit margins, restaurants need to adapt to rapidly altering circumstances and guidelines for health and safety, but it is simply not feasible to employ additional staff who can focus on monitoring CDC compliance at all times.

The Challenges

For restaurants to operate safely under COVID-19, they must strictly follow health code regulations for social distancing, mask use, personal protective equipment, and maximum occupancy limits in the front and back of the house. Any solution must be tailored with the flexibility to alter existing guidelines and parameters for social distancing and best practices as updated information becomes available.

Sarvanaa Bhavan’s existing camera system is not optimized for computer vision, so TensorIoT is customizing the equipment and programming calibration to support seamless integration. It is also critical that the solution include a method for getting a heatmap of high-traffic areas that require focused cleaning.

The Customer

Sarvanaa Bhavan is a restaurant chain who wanted to give customers the enjoyable restaurant experience they’ve been missing without compromising the safety of their customers or their employees, which is why they chose TensorIoT as their partner. TensorIoT’s SafetyVisor system is built from a foundational of industrial protection and custom tailored to help support businesses in safely operating under pandemic conditions. SafetyVisor provides enterprise grade performance, scalability and security in a low-cost package powered by AWS. The TensorIoT solution hooks into existing camera feeds and wireless sensors for data collection and monitoring. The lightweight solution does not require special training and can be deployed quickly.

The Solution

TensorIoT is expediting the process to incorporate existing camera hardware that’s readily accessible with fast and easy implementation of Safetyvisor. SafetyVisor has been deployed in the main Sarvanaa Bhavan restaurant location with the goal of implementing SafetyVisor across all of their restaurant locations to support social distancing, crowd control, and guided cleaning. TensorIoT is making sure to custom tailor the SafetyVisor solution to meet the restaurant’s needs through integrating cloud infrastructure with the ML model running on the edge.

Safetyvisor runs on multiple AWS systems, including Amazon Sagemaker and Amazon Lambda for ML processing, AWS Greengrass for functions running on the edge, DynamoDB to build the foundation of the cloud infrastructure for data processing and management, Amazon Cognito for rule-based access control, with SNS notifications for violations of the rules that have been given to the system.