How AI-Powered Video Analytics is used for Olympic Size Events
With the Tokyo 2020 Summer Olympics less than a year away, the venue operations and security managers are likely implementing and testing a variety of technologies to ensure the event runs safely and smoothly. Hosted on massive dedicated campuses, large scale global events like the Olympics, are operated much like a city: While the influx of residents and visitors is relatively temporary, the public safety, maintenance and operational management concerns of Olympic Villages and other event complexes are similar to those of a city. There are multiple athletic competition venues to secure, as well as a plethora of other properties, including parking lots and garages; concessions and restaurants; retail stores; and media and medical facilities. Each of these requires security on its own in addition to the overall event campus security.
Of course, the threat of a terror attack looms large over mass gatherings; however, the extensive property security, visitor safety and crime prevention preparations are often equally as significant as the anti-terror efforts, and there are many ways technologies can assist event organizers ensure world-class security readiness. There are biometric security screening solutions that can be leveraged for access control and identification; drone detection systems, such as those used in the 2018 South Korea Winter Olympics; and, of course, cybersecurity technology to address the digital threats to the event: Japan’s National Institute of Information and Communications Technology has already implemented an overarching cybersecurity strategy for the 2020 Olympics.
The most fundamental security technology leveraged by large sports venues and mass events is video surveillance. In fact, at large outdoor events, video surveillance systems have become a de facto standard, typically monitored by security staff in a command and control center where they monitor series of cameras in real time. The security cameras empower the on-duty police officers and act as a force multiplier to extend their oversight capabilities.
However, in practice, real-time monitoring is often impractical and unsustainable: With such extensive camera installations, it is virtually impossible to actively monitor every camera. Even if there were enough personnel to watch each video feed, human analysts are still prone to distraction and error, and critical details in video could easily be overlooked. Even when video is used post-incident during the investigation, the process of manually reviewing video footage demands extensive time and resources, alongside the same risk of human error and distraction. As a result, most video footage is not reviewed and available video evidence is review based on priority of what post-event investigators have time to comb to build their case.
To better leverage their video surveillance systems, mass event venues and stadiums are deploying AI-backed video content analytics software, which processes surveillance video; identifies the objects (such as people or vehicles) that appear in the video footage; and classifies and indexes the video object metadata, so that footage can be easily and quickly searched and analyzed. Here are some of the ways event organizers can leverage the technology to empower security teams:
Accelerate Investigations with Rapid Video Review
Based on Deep Learning and artificial intelligence techniques, video intelligence solutions enable users to filter real-time and pre-recorded footage for objects of interest with speed and precision, isolating in video the men, women, children, or vehicles, as well as their attributes such as appearance similarity, color, size, speed, path, direction, and dwelling duration. This is a critical capability in the aftermath of a security incident, because it empowers security agents or police officers to focus their video investigation on available intelligence about the incident. For instance, if witness accounts credit a man wearing blue and carrying a backpack with perpetrating a crime, law enforcement can review only the video that contains men matching that description, and thus reduce the video to review to the relevant footage. This forensic functionality enables investigative teams to quickly understand incidents and even identify persons, objects or behaviors of interest at the onset of an investigation, before there is any direction to the case.
Beyond isolating appearances of criminals and suspects in video, another key application for the technology at events is locating missing persons and reuniting them with their parties. Especially in cases where children and elderly visitors who have wandered off, making surveillance video searchable enables law enforcement to more easily find the missing persons in video feeds and even configure rules to trigger alerts when objects matching the person’s description are detected in real time.
Real-Time Alerts for Proactive Incident Response
By defining rules to describe persons, objects and behaviors that require security intervention, personnel can receive intelligent alerts based on video surveillance to drive immediate assessment and response to incidents unfolding at a mass event. For instance, video analytic systems can detect activity in marked sensitive areas; people dwelling beyond or number of people present in a pre-marked area exceeding the pre-defined “normal” threshold; or the appearance of specific faces or combination of object classes and attributes. By triggering alerts for these types of behaviors, security staff can assess whether someone is trying to access restricted areas; a person is suspiciously loitering; a crowd is forming; or a person of interest is on-site, and, ultimately, determine the best response to ensure the public safety of the guests and property.
Video intelligence alerts can be configured based on expected crowding figures. The operator can define the expected number of people, the area and the time duration which, when exceeded, requires security assessment. This allows event organizers to dynamically respond to growing crowds, whether it’s lengthening queues or people amassing in response to a security incident. Improved response to crowds also allows security teams to maintain pedestrian or vehicle flow, respond to emergencies and keep visitors safe and content.
Facial Recognition Alerts
To proactively prepare for potential threats, mass event organizers will often compile a watchlists of known criminals or suspected terrorists and monitor video feeds for appearances. Video content analysis streamlines this monitoring, especially those systems that include facial recognition capabilities. Operators can leverage face matching technology to find likely matches for persons of interest in video footage and also trigger alerts when faces are recognized in video. Once the operator affirms the face match, the security team can decide how to best track the person of interest to prevent incidents. With enhanced real-time situational awareness, security personnel can be dispatched quickly to diffuse threatening situations.
Versatile video investigation software empowers users to analyze video, investigate situations, prevent incidents and respond more quickly and efficiently to security events. Every two years the world witnesses at least one spectacle of sport that is fraught with potential security issues, but beyond the Olympic games, there are so many events that draw international attention and, with it, increased risk of security breaches, threats and incidents. Organizers can enhance security by leveraging game-changing technologies that extend their video surveillance investments. Moreover – as we’ll discuss in a future post – video content analytics offers organizers the ability to derive operational and business intelligence from video surveillance, further maximizing their video networks to optimize and streamline event performance beyond security and safety.
More on video analytics for hospitality and entertainment here.
Signup to receive a monthly blog digest.