“Persha Pekarnya” is one of the most popular bakeries in Odessa, Ukraine. The bakery sells and produces bread, cookies, cupcakes, cakes, pastries and pies, including a wide selection of organic produce for vegans and customers with gluten intolerance. “Persha Pekarnya” has three bakeries in Odessa, with a broader network of establishments across Ukraine.
“Persha Pekarnya” bakery was looking to protect their customers and workers in the retail areas of their establishments by enforcing strict epidemiological safety measures through mask-wearing and social distancing. They also wanted to control staff in the kitchens, to ensure they wear personal protective equipment while cooking. The desired solution had to be automated not to disturb customers, sales people and kitchen workers, and to be easily integrated with their existing security monitoring system.
VITech Lab customized their PPE monitoring solution to detect the required PPE items: face masks, gloves, white caps, and aprons. Face mask detection was a must for retail areas while other PPE had to be worn (and tracked in real-time) in the kitchens. The bakery’s safety monitoring system was reviewed and assessed for ease of integration with the PPE monitoring solution. The delivered solution was deployed on AWS to guarantee near-real performance; notifications were sent to the monitoring system’s UI for convenience.
VITech Lab implemented their PPE monitoring solution at “Persha Pekarnya” bakery in less than two weeks. The solution was enhanced to ensure PPE detection accuracy of 99% in retail areas and of ~97% in the kitchens. It was integrated with the bakery’s existing security monitoring solution, allowing them to avoid significant investment into new cameras. Since the solution was designed as an automated system, PPE violations by customers and staff were resolved quickly and without the unnecessary fuss, improving overall satisfaction.
Ukraine’s first patient with COVID-19 was hospitalized on March 3, 2020. Since then (as of October 13, 2020), 270 587 cases have been confirmed, with 5122 dead. To halt the spread of the deadly virus, a national quarantine was imposed on March 12, and it was only partially lifted on May 22.
Amidst the pandemic and strict quarantine measures enforced by the government, many businesses had to reinvent the way they work — serve customers, administer working spaces and shopping areas, manufacture products, and produce food & beverages. By law, they had to ensure that their customers wear face masks and keep social distance and that their staff comply with epidemic safety rules (e.g. wearing of gloves, change of face masks every 2-3 hours, etc.).
As a food-related business, “Persha Pekarnya” bakery did not have to close down during the national quarantine. For them it meant several things:The business had a good chance to survive, as opposed to companies that had to stop operations completely (e.g. gyms, clubs, shopping centers, etc.), or work from homeThey would have to undergo regular reviews by the officials, assessing their readiness to the pandemic; i.e. their compliance with epidemiological safety rules and regulations
“Persha Pekarnya” needed to act quickly to make sure their customers and workers were safe and to avoid government fines. They decided to automate their existing security monitoring solution to enable it to:Detect customers/workers not wearing a face mask in the bakery’s retail areasDetect kitchen staff not wearing face masks, gloves, aprons, and white caps
The company’s executive team researched the market and opted for VITech Lab, an established provider of AI/ML solutions around PPE compliance and monitoring.
The VITech Lab team’s first step was to research the problem that “Persha Pekarnya” bakery had; namely, they wanted to detect different types of PPE items (just face masks vs. gear for kitchen staff) in different spaces (shops’ retail areas vs. kitchen areas). It was obvious that, should they use VITech Lab’s PPE monitoring solution, customizations would be needed.
Then, the bakery’s security monitoring system was reviewed. The conclusion was that its cameras could be used to capture video streams for machine learning algorithms to process collected data in real-time. More cameras had to be mounted in the kitchen areas, though, to capture workers’ images from different angles.
Since the PPE monitoring solution was originally designed for laboratory safety, it could accurately detect lab coats, safety glasses, gloves, and masks. VITech Lab had to retrain the machine learning algorithm under the hood on new data — datasets featuring aprons and white caps in the kitchen environment. A new dataset featuring various face mask designs was also used to finetune the model.
Given the severity of the COVID-19 pandemic, VITech Lab acted quickly to deliver the solution for “Persha Pekarnya” bakery in less than two weeks. In total, the solution was implemented in the bakery's three locations in Odessa.
The solution proved to be highly accurate in detecting face masks in retail areas, regardless of the mask’s type (e.g. cloth, surgical, N95, etc.). It was also effective in figuring out if workers in the kitchens wear face masks, gloves, white caps, and aprons. The accuracy was 99% and almost 97%, respectively — a nearly perfect performance.
Since PPE compliance violations were detected automatically and acted on by specifically hired epidemiological safety workers, sales people could focus on serving customers instead of having to make them wear face masks. Knowing they are being “watched,” kitchen staff started to pay more attention to epidemiological safety regulations, too.
Automation of PPE monitoring through AI allowed “Persha Pekarnya” bakery to increase satisfaction of their customers and employees. Customers quickly got used to the fact that the bakery enforced mask-wearing and stopped complaining about it. Employees were quick to adopt stricter epidemiological safety measures as well. Overall, the solution allowed to take epidemic safety compliance to a new level, increasing safety in retail areas and in the kitchens to curb the spread of COVID-19.