Gone are the days of relying on word of mouth restaurant recommendations. Today, we have come to rely on Google reviews, or Zomato rankings to gauge the popularity of the newest hangout in town. Not only do star rankings tell us how swanky the place is, but also gives an indication of how good the grub is… and this includes hygiene levels too.

The F&B business in India, estimated to be valued at Rs. 30 lakh crore, has a lot at stake. The advent of social media has only made the landscape more competitive, as restaurateurs vie for prime reviews. Notably, maintaining the highest quality in standards and hygiene levels in kitchens is of utmost importance. This has created the need for enhanced activity recognition – a space that can be competently addressed by computer vision and machine learning.

Recognising The Need For Activity Recognition & Accurate Compliance Checks

Wobot.ai, a Delhi-based computer vision SaaS startup, founded by Tapan Dixit, Tanay Dixit and Adit Chhabra in 2018, specializes in utilizing the latest AI technologies for efficient activity recognition especially in the hospitality and retail sectors. COO and cofounder Tapan Dixit is extremely bullish about the surveillance and video analytics market’s growth prospects, as he credits the surge in internet connectivity levels like 5G that will permit sharing videos of high quality like 1080p; as well as high compute speeds for faster data processing on the edge.

“We realized the value in remote monitoring business operations, as most organisations were doing this manually. Physical monitoring of businesses is ineffective and expensive - the cost of monitoring compliance is around $500 per employees per year. Weekly or monthly audits are seldom done rigourously, and are also highly cost intensive activities,” explains Dixit.

So they had a two-step solution - first, different checklist items for various industries – like PPE, sanitization, hand-washing in hospitality were digitized, gaining granular insights into an industry’s workings. These checklists were converted into computer vision models for cameras. State-of-the-art neural networks and models form the backbone architecture of Wobot.ai’s platform that continuously learns and trains diverse use cases for Standard Operating Procedures (SOPs) for different industries. Secondly, the team saw huge value in utilizing an existing network of CCTV cameras (or possibly adding a few more in strategic locations in a shopfloor or restaurant kitchen) to monitor human activity to get the initial data stream. Wobot.ai has a plug-and-play offering that can be integrated easily with a CCTV network. The platform quickly identifies, documents, and distributes visual insights to stakeholders and decision-makers through real-time notifications via our Apps and user-friendly dashboards. The models have an accuracy rate of 95%. The platform has compliance checklists for various industries like retail, food, hospitality, kitchens, warehouses, manufacturing and healthcare.

Wobot.ai has over 10,000 installations and clients in India, the Middle East, and is expanding in the USA. Marquee customers include IRCTC, Akshayapatra Foundation, Rebel Foods, Airtel, CureFit, Travel Food Services, Burger Singh, Max Estates, Wow Momos, Apparel Group, and Kitopi, among others.

Pandemic Reinforced Need For Remote Quality Monitoring

When the pandemic hit in March, one of the widely impacted services was delivery – food, groceries, medicines and everyday essentials, over concerns of hygiene. Ecommerce companies scrambled to ramp up safety and hygiene measures to restore the faith of customers and keep their revenue margins. Food delivery apps like Zomato and Swiggy witnessed a major blow to their businesses early this year – Swiggy let go of 1500 employees and closed several cloud kitchens, while Zomato had to terminate employment of 500 staff. The food and beverage industry, for one, was undergoing one of its toughest times and safety has never been more paramount. Dixit says, “With the outbreak, the need to ensure hygiene, safety, security and premise sanitation has become paramount and with it, the need for remote monitoring has gone up. We do see the adaption of AI-based monitoring to continue as working models change in many ways. AI-based remote monitoring of operations takes the load off business owners and heads of locations and helps them in ensuring process compliance and improving the operations in their premises.”

When a human can monitor around 39,600 frames with concentration, Wobot.ai can monitor 6,48,000 frames in the same six-hour period. Video analytics tools like Wobot.ai can offer 24*7 monitoring and real time feedback on compliances and SOPs, reduces risk of non-compliance, brings down costs and secures employee safety.

Not Surveillance, But Qualitative Monitoring

One of the biggest challenges that Dixit and his team encountered was to explain to customers how the tool is not a replacement for manual tasks but supports to enhance manual tasks done by workers.

“If we look at the hospitality, manufacturing and industrial sectors together, the number of workers doing manual work would be above half a billion. Our endeavour is to support these workers in their daily manual tasks, not replace the tasks.” 

Dixit commonly refers to the tool as a “third eye”, and explaining that it should be viewed as someone giving feedback to the workforce, not a surveillance tool. “The tool not only spots anomalies but also shares positive feedback when compliances are met. To the workforce, we are giving this as a feedback and retraining tool to improve their performance.”  

Wobot.ai recently raised Pre Series A funding from Sequoia Capital and previously, a round of funding from Titan Capital.