Analyzing Culinary Success: Video Analytics in Hospitality

Video analytics allows businesses to make faster and better decisions, which helps improve different aspects of their operations. Here are a few ways that video analytics benefits businesses:

• Video analytics helps you save money and reduce inefficiencies: Many businesses have processes and resource usage that are not very efficient. With video analytics, you can identify these less-than-optimal processes and fix them. This saves your company both money and time.
• Video analytics improves quality control: You probably have ways to ensure that your services or products are of good quality. However, sometimes it's hard to find exactly where things went wrong if there's a problem. With video analytics, you can track and analyze your business process from start to finish. This helps you consistently deliver high-quality services or products to your customers.
• Video analytics boosts marketing results: You can't read your customers' minds, but you can study their behavior to understand what they like and don't like. When you truly understand your customers, you can make better offers and improve your marketing efforts.
• Video analytics enhances customer experience: By improving your marketing, processes, services and products, you create a great experience for your customers with fewer hassles. A great customer experience leads to customer loyalty, strong brand support, repeat purchases, and increased sales.
• Video analytics gives you instant insights: When you have a video analysis system in place, it automatically adds new data. The algorithms immediately analyze this new data and provide you with valuable information in real time. This helps you make quick decisions and transform your business into a nimble company.
• Video analytics gives you a competitive advantage: video analytics provides faster and more accurate information. With these insights, you can act quickly, increase your profit margin, and use your resources more effectively than your competitors. Even if your competitors are also using video analytics, you'll be able to keep pace and maintain your market share.

First in this post we'll look at a set of basic events that can be extracted from video data. Then we customize this basic set for a specific business, in this case, a restaurant. And we will demonstrate some analytics capabilities based on the data obtained.

Basic set of detected video events

The simplest set of basic events for our purpose might include, for example, the following events.

• Event 1: The appearance of a new person in the frame.
• Event 2: The disappearance of a person from the frame.
• Event 3: A person moving at high speed in the frame.
• Event 4: Сamera being blocked by an object.

Let's look at examples of less trivial events.

• Event 5: A person passing through a specific area in the frame.

This event can be used, for example, to monitor doors and corridors.

An example of control of passage along a park path *.

• Event 6: A person or group of people being motionless for a certain period.
• Event 7: A person approaching a group of people.
• Event 8: A person walking away from a group of people.

These three events are illustrated in Figure.

An example of a stationary group of people, as well as a person walking away and then approaching this group.

• Event 9: A person entering a defined area in the frame and remaining there for a while.
• Event 10: A person walking out of a specified area in the frame after staying there for some time.

This pair of events can be used to monitor specific areas of the frame.

An example of two people stopping in a marked red zone on the sidewalk.

Let's now discuss how the above basic events can be customized to satisfy the needs of a specific business, in this example: to monitor the activities of a restaurant.

Restaurant — customized video events set

By customizing the events set in video analytics, businesses can personalize the system to focus on events or activities that are directly relevant to their operations.

Customizing the events set in video analytics allows businesses to align the system with their unique operational requirements. By defining specific events to monitor, companies can gain valuable insights into their workflows, identify bottlenecks, and streamline processes.

With customized event detection, businesses can obtain tailored insights that fit with their specific goals and objectives. By specifying events that are highly relevant to their industry or business model, companies can extract valuable information on customer behavior and market trends.

Event 5 (passing through a specific area) could be used, for example, to count visitors to a restaurant.

Event 6 (a person is motionless for some time) allows you to monitor the occupancy of tables in a restaurant. The results of such monitoring can be presented in the form of a visual diagram.

If we add detection of events 7 (a person approaching a group of people) and 8 (a person walking away from a group of people) to event 6, then it becomes possible to monitor the work of waiters who serve tables. Then the previous diagram can be supplemented accordingly.

Similarly, the diagram can be supplemented with other features - from the presence of a hookah on the tables to the number of bottles on them.

Using events 9 (a person entering a defined area in the frame and remaining there for a while) and 10 (a person walking out of a specified area in the frame after staying there for some time) allows you to monitor, for example, a bar counter.

Diagram of the occupancy of the bar counter and the corresponding work shifts of bartenders.

Conclusion

Businesses are interested in using video analytics to make more money, provide better customer service, and find and fix problems. By studying hidden patterns in video data, businesses can discover valuable insights that help them make smarter decisions, predict future trends, get rid of inefficiencies, take advantage of customer buying habits, and improve their marketing.

In this post, we customized a basic set of video events for just one specific business - a coffee shop. However, similar customization of the basic set of events can be done for other businesses. For example, for entertainment park, retail stores, warehouses, parking lots, construction sites.

The algorithms developed by LettiNotes have the capability to detect and track objects in the frame. They also consider the relationships between events and the context of the current situation. These algorithms can effectively process low-resolution videos captured in low light conditions. This addresses both privacy requirements and the typical lighting conditions in the environments we serve.

* S. J. Blunasden, R. B. Fisher, "The BEHAVE video dataset: ground truthed video for multi-person behavior classification" , Annals of the BMVA, Vol 2010(4), pp 1-12.