AI AND VIDEO ANALYTICS BLOG
Video Surveillance & Physical Security Industry Viewpoints
July 29th, 2020
Author: Lizzi Goldmeier

Optimizing Campus Safety and Security in a Period of Political Unrest and a Pandemic

Video content analytics helps ensure public health and safety

The combination of current political tensions and COVID-19 concerns will add significant challenges for college administrators and security teams for the 2020-2021 academic year. Ensuring the health and safety of everyone on campus — students, staff, and faculty — is always a high priority, but during an election season and COVID pandemic, the health and public safety aspects of residential and academic life pose additional hurdles and risks.

Mitigating health safety risks on-campus

Whether students will return to most college campuses for the fall is questionable. To avoid the spread of COVID-19, it is clear that many higher education institutions will conduct courses exclusively online. However, depending on the local and regional status of the pandemic, some colleges may decide to try to hold classes and events on campus. Furthermore, some universities will conduct other business: some administrative staff will continue to work, and researchers will carry on their projects. Either way, the campuses will not be entirely vacant, yet the scene will be anything but “business as usual.”

Typically, campus public safety officers prevent, respond to, and investigate incidents such as theft, assault, vandalism, as well as hazards that could cause fire or flooding in a building. And now, with COVID-19 threatening the well-being of everyone on campus, security teams will also play a role in at least monitoring how well people on campus comply with health safety mandates such as wearing face masks and physical distancing. University health services, too, will have additional responsibilities, in terms of contact tracing.

Ensuring safety during political protests

Another consideration when it comes to reducing crowding, virus spread and violence is protests – which certainly are not a new threat to college campuses, but which carry new challenges with the changing public health standards. The spring and summer of 2020 saw many protests across the United States and elsewhere; it’s possible this trend will continue or re-emerge in the fall/winter. Another factor to consider in the United States is that a presidential election will be held in November; this also could spur some in-person political protests, whether for or against a cause or a candidate.

Video intelligence software as a force multiplier

Neither health services staff nor campus police officers can be everywhere at once, even in normal times; how can they take on these additional tasks, without expanding their staff or breaking their budgets? The answer lies in leveraging video surveillance networks, which most universities already have in place. CCTV video cameras enable security staff to monitor situations in real-time, or review video footage after an incident. By pairing this video surveillance infrastructure with video content analytics software, universities can get valuable data from their video footage to combat crime on campus, prevent the spread of COVID, and more. Powered by deep learning and artificial intelligence, video content analysis enables complex object extraction, recognition, classification, and indexing activities, thus making video searchable, actionable and quantifiable.

Investigate incidents more quickly and effectively

In the aftermath of an incident it is time-consuming for security staff to manually sift through footage from various cameras to glean information, and the process cannot rule out human error.

Fortunately, video analytics software makes it possible to conduct a forensic review of footage, based on object classifications, such as cars, trucks, buses, motorcycles, bicycles, women, men, children, and animals. The speed, paths, direction, and dwell durations of objects can also be detected. System operators can search and filter objects based on their classes and attributes, including face recognition, license plate recognition, appearance similarity, vehicle or person clothing color, and size. This technology helps campus security teams accelerate investigations by reviewing incident footage in minutes rather than hours or days.

To give another example, if a student is reported missing, campus police can search footage from multiple cameras across campus, filtering based on appearance similarity, such as  a female wearing red upper wear, blue lower wear, walking due west, between the hours of 9 pm -1 am. The technology also can be used to search and filter by face recognition, seeking a match for a video or still photo of the student, to track the most recent appearance of the student on campus.

Face recognition and license plate recognition alerts

Video content analysis also enables rule-based, real-time alerts to help increase situational awareness and improve response times. In the case of a missing student, face recognition can allow campus police to detect the individual in real-time based on an uploaded image or video capture of the student and configuring an alert to be triggered whenever a potential face match is detected. A human operator can then evaluate the accuracy of the match and determine next steps for recovering the missing person.

In addition, face and license plate recognition can be used based on watchlist configurations to alert when individuals or vehicles either included or excluded from watchlists are detected in certain spaces. During campus protests, this can be leveraged for tracking past agitators and proactively preventing outbreaks or for more closely monitoring unrecognized vehicles that enter certain areas of the campus. This enables security to keep a closer eye on suspicious or unauthorized people and vehicles and respond proactively to situations, if they develop.

Real-time alerts based on behaviors and activity

Once system operators have established normal campus behavioral patterns video analytics systems can help operators detect anomalous behavior or objects, such as illumination changes. For example, if a light is turned on or off in a classroom, gymnasium, or office during a pre-selected time frame – when activity is unexpected in the facility – alerts can be triggered to enable the surveillance operator to quickly assess the situation and decide whether to dispatch a security guard to investigate whether theft or vandalism is in progres

Another behavior that can be detected by intelligent video surveillance is dwelling – systems operators can set thresholds for normal dwell durations which, when exceeded, alert security. For some university campuses, theft is a major problem, and lingering for an unusually long time can sometimes indicate an intention to commit a crime. It may also indicate that someone is experiencing a medical event and needs emergency care. In either case, the dwell alerts can help security assess and respond proactively.

Measuring face mask compliance

In light of the COVID 19 pandemic, another behavior that campus security may opt to track through real-time alerting is face mask wearing. Over time, however, intelligent video surveillance can be used to aggregate data and gather anonymous statistics regarding compliance with health safety recommendations and mandates, such as how often and where people on campus are wearing face masks. This information can then be evaluated and analyzed to determine any correlation with infection rates (or lack thereof) on campus.

Enforcing social distancing and reducing crowding

Similarly, with proximity identification capabilities, administrators can leverage video analytics to measure, quantify and analyze the distances between individuals for reporting purposes. Generating dashboard visualizations and reports of physical distancing data — across campus, over time — enable campus administrators to determine whether staff, students and visitors are generally complying with the physical distancing mandate and identify problem hotspots where (and when) violations tend to occur. Based on this data, administrators can understand where and when pedestrian crowding and congestion tends to occur, so the university can adjust signage or traffic flows as needed and improve infrastructure even for when physical distancing is unnecessary.

To maintain physical distancing, universities may limit the number of people in a building, such as a lecture hall or gymnasium. Intelligent video surveillance can help keep track of building occupancy, tracking movement across multiple entrances and exits in a facility. Alerts can even be triggered when occupancy thresholds are violated in real-time.

People-counting alerts can also be configured to track and prevent crowd formation, by triggering notifications if the number of people in an area exceeds a pre-determined threshold. When alerted, security can assess the situation and determine how to respond. Coupled with face mask detection, this capability enables security to uncover, when crowds do form, whether the individuals are wearing face masks and physically distancing. For campus protests and even everyday traffic, this empowers proactive prevention efforts.

Facilitating contact tracing

If a student or staff member tests positive for COVID-19, the college health service will want to notify any students, staff or faculty who had substantial contact with the infected individual. The health service can use a combination of the face recognition, appearance similarity, and proximity identification functions of a video content analysis system to analyze video footage to find out where that individual was on campus and with whom the individual interacted. While maintaining the anonymity of the infected individual, administrators can prevent further spread by recommending self-quarantine to the students or employees who were exposed. Meanwhile, those who were not detected near a diagnosed person can be more at ease and will not have to self-quarantine.

Gather long-term trend data for business intelligence

In addition to real-time notifications, video content analysis also delivers aggregated data collected over time, via customized reports and visual dashboards that include custom graphs, heatmaps, and histograms. For example, administrators and security staff often want to see vehicle and pedestrian traffic flows, so they can understand which areas of a campus (buildings, parking lots, or roadways) commonly experience traffic congestion and bottlenecks. This quantitative data helps college campus administrators and security teams research trends so they can make better decisions based on statistics rather than human observation, to prevent future security problems. This results in less traffic, less crowding, and a safer environment for all.

College campuses can find ways to rise to the occasion of these unprecedented and difficult times, and optimize their security operations. By pairing their existing video surveillance technology with video content analytics, college administrators, security teams, and health officials have a comprehensive toolkit for investigating crimes, preventing health and safety threats, and providing long-term business intelligence for optimizing efficiencies beyond the pandemic era.