Best Practices: Using Video Content Analytics to Enhance Security, Health, Safety and Efficiency in Mass Transit Systems

Nov. 19, 2020
By enhancing existing video surveillance infrastructure with complementary video content analytics technology, agencies can proactively respond to situations, uncover trends and more.

How can mass transit agencies optimize their security and operations departments during a period when ridership revenues are dramatically lower due to the COVID-19 pandemic and, as a result, state and municipal budgets are also suffering mass transit funding cuts? Mass transit agencies already must juggle numerous priorities of commuters, taxpayers and local officials: frequent, on-time service; sufficient passenger capacity; safety; and cleanliness.

To accomplish those goals – in response to the pandemic and otherwise – transit agencies must monitor miles of track, dozens of depots, hundreds of platforms and millions of commuters. Video surveillance makes it possible to oversee these properties, but while those camera networks are helpful, they are insufficient on their own. Transit security staff could not possibly efficiently and accurately monitor every camera in real-time or manually review video footage for post-incident investigations.

To overcome these challenges, transit agencies can enhance existing video surveillance infrastructure with complementary video content analytics technology. Powered by deep learning and artificial intelligence, video content analytics technology extracts, identifies, classifies and indexes valuable video metadata to make it actionable, searchable and quantifiable. Operators can configure alerts based on object classification and recognition to proactively respond to developing situations, productively review footage and uncover trends by analyzing aggregated video data.

Increase situational awareness for proactive prevention

The more transit operators are aware of what is happening in their environment in real-time, the more easily they can respond to a developing situation; this applies to customer service agents as well as security officers. Video content analytics systems can be configured to detect objects and behaviors based on pre-defined settings. Based on traffic norms and benchmarks, alerts can be configured to detect unusual or unexpected behaviors and issue notifications in real time. For example, if there is a person detected in a restricted area, the system can send a real-time alert to security staff who can assess whether the person is authorized to be in that area. Similarly, people-counting alerts can be used to indicate when crowds or queues form on platforms or at ticket counters, so operational or customer service staff can proactively intervene and provide a positive, streamlined visitor experience.

Filter results with face recognition, license plate recognition and appearance similarity

Another important application for alerting is finding persons of interest — whether a missing child or a suspicious individual. Without intelligent analytics, many surveillance operators simply cannot drive proactive and preventative responses in real time and they rely on post event investigation to track individuals of interest. Even this can be difficult without video analytics software – especially when there are hundreds or thousands of commuters in a subway or train station. With traditional surveillance, security staff search by foot and vehicle patrol, monitor video feeds manually in real time or sift through volumes of archived video footage. Video content analysis offers powerful search and filter capabilities that empower security teams to focus video search based on object classifications and attributes, such as clothing color, speed and directional path. Filtering video helps accelerate searches and allows investigators to pinpoint specific people, vehicles or other objects with speed and precision, saving post-event investigators valuable time and enabling them to quickly resolve situations and solve cases.

These same filters can also be leveraged by security teams to prevent further crimes by developing real-time alerts based on past incidents. Whereas in a post-event investigation an officer would search for suspects based on clothing or vehicle filters, in real time, surveillance operators could create alerts to be notified when individuals matching suspect descriptions are detected, allowing for actionable intervention in the immediate aftermath of an incident.

In jurisdictions where it is allowed, investigators can use facial recognition to search video and detect when specific individuals appear in a live camera view. Operators can create a watchlist of digital images or videos of persons of interest so that a video analytics system can match those images against faces that appear across multiple cameras; when the face is detected the system sends a real-time alert to security managers, who can then review and confirm the match before deciding on and executing a real-time response.

It’s important to remember that facial recognition is not a video investigator’s only recourse: Operators may create a real-time alert based on witness descriptions of the individual’s clothing, bag or hat, as well as associated vehicles and pinpoint possible matches based on appearance of similar features.

Similar to face recognition, security officers can search and alert on video footage for a license plate number that is connected to an incident. If they place a complete or partial license plate number into a watchlist, they will receive a real-time alert if a matching or similar license plate has appeared in a video camera view.

Improve the visitor experience and health safety compliance

While the COVID-19 pandemic presents additional challenges, offering clean, safe and uncrowded transit options are key performance metrics of every bus, train, trolley and subway operator, even in the best of times. Currently, to prevent the spread of infection, some port authorities and states have mandated that riders and staff wear face masks and practice physical distancing. When video content analytics systems are trained to count objects, recognize whether people on-site are or aren’t wearing masks and even measure the proximity between people, managers and security personnel can configure alerts to notify staff so that they can draw conclusions from aggregated, anonymous video data as to whether people are complying with mandates. These capabilities also enable management teams to prove compliance with face mask and physical distancing mandates, even if only among their employees. By using a video analytics system to aggregate long-term data over a period of days, weeks or months, managers can see where and when employees and commuters are complying with the mandates and decide whether operations or staff changes are needed to support compliance.

By monitoring the proximity between people to promote physical distancing and prevent crowds, operators improve the customer experience. Operators can also combine people-counting, proximity and other object classification analytics to drive reporting and alerting to understand crowding patterns, distance violations and public safety trends.

Video content analytics can be leveraged to monitor the number of persons in a building in real time and over a period of time. Systems operators can set up custom people count alerts and line-crossing metrics to quantify the number of persons who have exited and entered a building and can even aggregate the total numbers from multiple cameras across numerous exits and entrances to determine if a building has reached excess occupancy.  

This is important not only for monitoring and maintaining safe building occupancy and preventing crowding, but also for purposes of disinfecting subway or train cars, as well as restroom facilities. As a fundamental part of everyday operations in bus terminals, train stations and airport terminals, maintenance is typically scheduled according to a clock, rather than on actual usage. Maintenance managers could more effectively deploy staff and clean properties by cleaning based on actual usage, by tracking the number of people who entered a restroom; notifying managers in real-time when thresholds have been met and assigning cleaning responsibilities based on actionable and quantifiable video data.

Gather long-term ridership data

Video analytics software can provide historical quantifiable data and dashboard reports to demonstrate key metrics, such as which transit stations are most utilized or when peak or lull traffic periods occur. For instance, anonymous visitor footfall trend data can empower management teams to consider how to best deploy staff, change exits/entrances, change transit schedules or change signage in some areas. Planners and marketing departments can also gather demographic data about their ridership, such as the percentage of men, women and children who pass through various transit services or stations. These aggregated data reports help transit management teams make better decisions about scheduling, staffing, construction, services, marketing and more. They can better communicate the value of their agencies because they can provide meaningful, data-rich reports to their constituents and state, local and federal authorities.

It is worth noting that comprehensive video content analytics systems that offer fully integrated facial recognition can also drive reporting about unique traffic and exclude employees from these analytics reports. By anonymizing the facial recognition data, while identifying unique visitor identities across cameras and days, facial re-identification empowers managers to drill deeper into visitor and employee activity and develop plans for promoting safe and healthy practices in public transit.

In an era of tight budgets and low ridership, transit agencies must find ways to get more value out of their existing technology investments. Video intelligence software is a key way to maximize their investments in surveillance networks because it accelerates investigations and video search; increases situational awareness with real-time alerts; and enables data-driven planning and decision making. With such technology, mass transit agencies can optimize their operations, improve their planning and ensure a safer, more pleasant riding experience for their constituents.


Stephanie Weagle serves as chief marketing officer of BriefCam and leads the company’s global marketing initiatives and accelerates market adoption of its video analytics solutions.

About the Author

Stephanie Weagle | Chief Marketing Officer

Stephanie Weagle serves as Chief Marketing Officer of BriefCam and leads the company’s global marketing initiatives and accelerates market adoption of its industry-leading video analytics solutions. Before joining BriefCam, Weagle was vice president of Marketing for Corero Network Security, where she led global marketing for the company’s cyber-threat mitigation product portfolio. Previously, Weagle held senior marketing roles at Lionbridge Technologies and Novell, Inc.