Computer Vision in Convention Centers: 9 Use Cases

Practically every industry has changed due to COVID, but the events industry underwent arguably the greatest transformation of all. As an industry that relied almost entirely on large, in-person gatherings pre-COVID, the events industry took a major hit when lockdowns were put into place. 

Not to be beaten by the virus, event planners pivoted to virtual events: instead of gathering face to face, which inherently restricts potential attendees to only those who are either local or can afford to travel, event organizers hosted their events online so that people around the world could join in from wherever they were. 

In addition to improved accessibility, event organizers realized there was another benefit to virtual events: data collection. Although event operators had been making strides with their data collection initiatives before COVID, hosting events online allowed organizers to collect more precise data than ever before. 

Now that in-person events have resumed, event organizers are looking for ways to continue collecting similar data that online events generated even as the industry shifts back towards in-person and hybrid events. 

Computer vision is a technology that can help event organizers and venue operators gain a similar level of insight into their operations. With guest journey tracing, real-time occupancy tracking, staff detection, and more, operators can more easily bridge the data gap between hybrid, virtual, and in-person events. 

What Is Computer Vision?

Computer vision (CV) is an emerging technology that uses artificial intelligence (AI) to interpret and draw conclusions about the visual world. It’s being used in industries as diverse as retail, construction, restaurants, and of course, events. 

The precise implementation of CV tech varies from industry to industry, but it always relies on some core fundamentals. First, a CV model must be provided with visual data. This can be a photo, video, live video stream, or other form of visual media. In commercial applications, like Safari AI, a venue’s pre-existing security cameras can be used as the data source. 

From there, the model will utilize deep learning, machine learning, neural networks, and other types of algorithms and AI models to parse the data into a form it can understand. This typically means breaking it down into simpler geometries in a process called object recognition. 

Once the visual data has been translated into the model’s “language,” the model will employ another set of algorithms to figure out the relationships between the different objects it has recognized. At this point, the model can start to draw useful insights about what it’s seeing. 

A quick example will help to clarify. Imagine that a CV model is deployed at a convention center to determine how many guests are attending a specific booth. The model will be fed a live stream of the data and use a set of rules (an algorithm) to discern people from booths. Then, it will use another set of rules to determine when a person approaches the booth. Finally, it can report the count back to the event organizer for human interpretation. 

9 Use Cases for Computer Vision in Convention Centers

Computer vision is a quickly evolving field and new applications for this technology are constantly being discovered. Currently, computer vision can be used in convention centers for:

• Guest/Door counts
• Guest Journey Tracing
• Dwell time
• Real-time Occupancy Tracking
• Throughput
• Live Wait (Queue) Time
• Heatmapping Asset Utilization
• Parking, Curbside, and Loading Dock Management
• Staff Detection

1. Guest/Door Counts

Guest counts, also referred to as door counts, are used to measure the number of people entering a space through a specific entrance. Although events can typically get similar data from simply measuring ticket collections, door counts have some additional uses.

For one, comparing door count data with ticket collection data can alert operators to potential security issues — if the CV model detects more visitors entering than tickets collected or passes scanned, that can indicate either a physical problem with security barriers or understaffing issues. 

For free, public events, the benefits are more obvious: convention center operators and event organizers can get a clear idea of how many people are entering the venue despite no tickets being collected. 

2. Guest Journey Tracing

Guest journey tracing allows event organizers and venue operators to get a better idea of the path that guests take through a convention center. By using object recognition algorithms to detect people, a CV model trained in guest journey tracing can capture data about the paths that individual attendees take. Then, it can aggregate that data to draw conclusions about the average paths that guests travel through the venue. 

For example: the CV model could follow the journeys of 1,000 guests to find that 40% go straight after entering the venue while 20% go right and 40% go left. Venue operators can then use this data to set premiums on booth space in highly trafficked areas or to evaluate ways to redirect flow more equally through the space. Vendors can use this to figure out the best spots to place their booths. 

Of course, real world models can do more than just determine whether a guest typically goes right or left. Complex models can follow individual guests through the entirety of their journey within the convention center so that operators can get data about the various stages of their experience — perhaps the average guest visits ten booths before getting a bite to eat and then visits another 20 before leaving the event. 

This type of granular data can help organizers make precise and detailed adjustments that can improve the guest experience and boost attendee engagement. 

3. Dwell Time

Dwell time refers to the amount of time that a person or object spends at a location. Event organizers can use this metric as a proxy for engagement — instead of only using guest journey tracing to determine where guests are moving in the space, dwell time can provide insight into how long guests spend at a particular booth or location, which likely indicates how engaging they found the exhibitor. 

On the other hand, if there are long dwell times in areas that are not meant to be engaging, that can indicate that guests are confused about where to go or that there is a traffic bottleneck.

With this data in hand, convention centers can make better strategic decisions about booth placement and content as well as optimize the flow of the crowd.

4. Real Time Occupancy Tracking

Although the era of social distancing is coming to an end, making sure that event venues aren’t overcrowded is still of the utmost importance — all events must comply with local fire regulations. Filling a convention center over capacity is dangerous and can lead to tragedy. 

Accurately gauging how many people are in a space has historically been difficult to do. However, new computer vision models can use live streaming video to monitor occupancy in real time and send alerts whenever occupancy limits are being neared or exceeded. From there, staff can take appropriate action, such as restricting entrances or removing guests. 

5. Throughput

Throughput refers to the amount of people moving through a given area over a specified time period. This metric is similar to dwell time and guest journey tracing but with a greater emphasis on the volume or rate of people passing through an area. Consequently, convention center operators can use this measure to make better decisions about where to place the biggest booths, which are designed to serve the greatest number of attendees. 

6. Live Wait (Queue Time)

Lines are undoubtedly one of the least enjoyable aspects of any event. But the worst kind of line is one that you don’t know how long you’ll be waiting on. To keep attendees happy, it’s crucial to manage their expectations: if a line moves faster than expected, they’ll be content, but if it takes longer than expected, they can quickly become irate. And if they have no idea how long it will take, they may just not even get in line in the first place. 

By combining real-time visual data with historical data, computer vision can help keep your attendees satisfied by providing them with more accurate wait time estimates. This can improve engagement by reducing the intimidation factor of long lines as well as increase guest satisfaction.

On the operator side, queue time estimates can help event organizers to find bottlenecks and points of inefficiency, and to better understand operational flow. 

7. Heatmapping 

Heatmapping is something of a hybrid between dwell time and guest journey tracing: it provides a visual representation of where guests are spending the most time in the convention center and/or what resources they’re using.

When heatmapping is used to indicate what spatial locations guests are occupying, the heatmap would display colors to represent the areas of the floorspace that get the most traffic. When heatmapping asset utilization, the resources themselves (booths, for example) would be color coded to indicate utilization (how many people take a flyer, for example). 

In both cases, heatmaps can be used to make strategic decisions concerning booth placement and crowd flow. When spatial heatmaps are compared with asset utilization heatmaps, they can give insight into underutilized booths — if there’s a hot spot in front of a booth but the booth is a cold spot on the asset utilization heatmap, this may indicate that the booth should be placed elsewhere or needs some other changes (assuming no errors in the model or data collection). 

8. Parking, Curbside, and Loading Dock Management

Although event organizers are mainly focused on what goes on inside a convention center, venue operators need to be cognizant of the logistical factors that can sink an event before anyone even gets in the building. 

Parking is one of those factors: it can be extremely aggravating at large events, and venue operators need to be vigilant to ensure good traffic flow and easy access to the venue. With CV, venue operators can easily determine the occupancy of parking lots and use that data to improve the guest parking experience. For example, screens that display real-time occupancy can be set up outside of various sections of the lot so that guests can more easily decide which areas to try to get a spot. 

9. Staff Detection

Since COVID began, staffing shortages have been a widespread issue. When resources are scarce, managers need to carefully allocate them so that they can be used as effectively and efficiently as possible. 

Computer vision allows event organizers to automatically detect staffing levels at different booths and exhibits within a venue. When the model notices that there are too few staff for the number of people currently at a booth, a manager can step in and redirect staff to that area. Similarly, if there are too many staff at a relatively empty spot, they can be rerouted to spaces where they can be utilized more effectively. Over the long term, organizers can use this data to be proactive and better decide where to put staff in the first place to avoid bottlenecks.

Key Takeaways: Computer Vision in Convention Centers

As event planners move back to in-person events, there is an increasing need for high-quality data that rivals that which can be collected from virtual events. Computer vision, a subfield of AI that uses complex algorithms to interpret visual data, can help event planners and venue operators collect accurate data and make better strategic decisions. 

Learn more about occupancy monitoring.